📚┃suno-school

[half dragon]

It's still there for me. There's also a Edit button by Publish button on the song listing.

[fair adder]

well it's not anymore for me, strangely. or at least my screen reader doesn't read it anymore. mmh... 🤔

[polar spade]

Thanks for asking. Occasionally, folks are happy to give a listen when things are slow. Just don't make it the only reason to visit a room. 🙂

Post the song here and let us know what is the purpose for writing the song, as in, what are you wishing to do with it.

[remote ivy]

feels like it worked here in the instrumental break: https://suno.com/s/kmZwwq6FkyeS0sN8 w/ as simple as tags as I put

unrelated, has anyone reported that excessive new lines between sections causes the song to replay the start of the song instead of continue to next section?

[remote ivy]

well i mean not the bar lengths, nor the saxophone hopping in after lucky, make it sing..but good enough

[frozen dagger]

Curious what screen you are on (Create tab, Library etc), maybe an option was changed. From what I can see so far it looks like it is in the same place in most places i am looking so far. When I click the 3 dots, Remix/Edit is the first option, hovering over that reveals more choices, the first one is Open in Studio, the second option is Open in Editor. Then Cover and more options after. Not sure if telling you where is is located will help. I hope you figure it out.

[lean bolt]

I need a little help everybody.

I am losing alot of credits trying to extend my song. I have extended from the point I want to, and removed all the lyrics above that spot.

The song keeps reverting back to the beginning instead of starting at the point I am telling it to.

Any suggestions?

[polar spade]

What are your sliders set to? Default?

[lean bolt]

nah

0 / 100 / 0

[polar spade]

Have you started any of the extensions at default?

[lean bolt]

not sure why I would do that, so no

[polar spade]

Try default settings first then. Start with a benchmark.

[lean bolt]

but... I just basically switched from extend to something else, then shifted back to extend and now it worked? so... maybe it was just a glitch?

[polar spade]

OK. Lets go at this another eay then. If it helps. What do the extension that start from the beginning sound like. Do they go further than the original song did?

[lean bolt]

HUH?

That made no sense to me

[polar spade]

OK. we'll call it a glitch then. Glad it worked.

[polar spade]

Ill try to say another way.

Ytou said the extensions started from the beginning. If your original song was 3:00, and you extended, and it started over from the beginning, how much of the beginning did it follow through? Did it sing the whole original song and then add more?

Did ytour extension add another 3 minutes? Likely not.
So how long then were these extensions which started over from the beginning?

[lean bolt]

oh no idea because I just moved passed them since they didn't start at the right place... but they were going to the length of full songs.

[fiery ruin]

Has anybody had difficulties over the last week with SUNO producing music close to what they've even asked for?

It seems like I'm having a much more difficult time getting what I asked for even with using previous prompts that I did a couple of weeks ago that worked perfectly fine.

Any help appreciated, please tag me.

[polar spade]

If it happens again, and they are a good version, you can skip the original and try to extend from the nice extension. Suno does extend from extensions differently than it extends from originals.

But thats just a extra tool to try if it happens again. Glad things worked then.

[lean bolt]

oh it wasn't an extension from an extension.

[polar spade]

Nothing to worry about. Just an extra technique to use. Keep doing what youre doing then.

[remote ivy]

dang near ever render keeps restarting the song instead of doing the final chorus....this has been a common issue for at least 2 weeks for me and never an issue before

[polar spade]

Its pretty wide open to guess at why certain prompts work differently today than last week. Even between generations, the same prompts can have a wide range of results.

[remote ivy]

i don't even listen to renders anymore i just skip to 2/3 of the way if gen time looks long in the tooth to confirm i got hosed and just wasted 10 precious credits

[fiery ruin]

I understand there's a certain level of weirdness and that's fine as I've tuned them and the prompting style to try to get as close to what I've asked for as possible.

I never realized how horribly difficult it would be just to get one single solo instrument and it's turned out to be in a complete and total nightmare...

The other thing I have encountered over the last week is that whatever they've done to it, and now completely ignores the negative prompting altogether.

[polar spade]

It may be a longer way around, but have you tagged a few places where we evetually get a solo anywhere that can be spliced and pasted where we want it?

[fiery ruin]

I'm not quite sure I understand what you mean. I am by no expert and how to use SUNO, but I have managed to get a few things to do sound good.

I've tried different combinations with both the lyrics and styles sections along with using the negative prompt sections to try to reinforce what I didn't want. This last week really has changed The stability of the product altogether.

[polar spade]

Ahhh.... im assuming youve used /had practice with the Edit UI as well.

Technically, ver 5.0 is out of beta and there should be no "updates". So the exact prompts, exact excludes and ..... well, if nothing is touched and we do a Cover of the same song from last week, it should have a similar vibe.

Even a change in lyrics can add something to the strength of style. So, what are you doing to want to use the same prompts, mostly? Trying to get a new song to have the same feel, tempo, instrumentation as a previous?

[fiery ruin]

I've had tried using the same prompts just so I could try to rule out anything I was using to see if there was a difference in terms of similarity. Even with using the exact same prompt and simply regenerating, what's produced is completely unrelated in any way to what was previously produced. Previously I would ask for no vocals in the negative prompting and it would do that just fine. Now it completely ignores it even with the same identical prompt.

One of the biggest concerns I was working on was trying to build a reproducible pattern that would give me a certain direction that I could tweak and tune to get small changes which had previously worked fine.

[fair adder]

i tried both on the library and workspaces, and for me the first option under remix/edit is cover. i don't know why

[polar spade]

I get that, but I have seen Suno take a left turn often on similar things. But still, the same in prompts and such should get a similar vibe in most instances. If its continually, with no alterations between last week and this, nothing more to suggest than to report it as an anomaly in the bug channel.

So , youve used Suno for a year or so? What's your experience then on working with the sliders? If any adjustments there that can help, even from something done last week.

Additionally, once we have a "vibe" we like, its helpful to make a persona. Then when using that persona on a new song, we choose the "legacy" version instead of the "regular" or vocal/Vox version. This gets us both vocal quality AND Similar instrumental vibe to show up in the next song we create.

[frozen dagger]

the only thing i can think of is to show someone else the screen? i'm checking my alt account for Suno as it has some different UI stuff, but the Editor is still second choice from Remix.
I presume you already knew to wait for the song to fully load? If the song was not fully finished being created the option for editor would be greyed out. But even then I think your screen reader would read that to you.

[fair adder]

it did before the ui changes a few days ago. i'll record a video recording with my speech viewer active so i can show it to the team. probably will do it tomorrow though, it's getting late where i am

[frozen dagger]

i used control+ win key + enter to toggle my windows screen reader on to check that as well. Not sure what you use, but just wanted to see if it was an obvious thing, but it seems to be working correctly.

[fiery ruin]

Prior to this week, the sliders worked very well and did do what I wanted. This last week they have seen to be completely useless.

I don't know if you can call what I do a vibe versus a specific type of genre. That in itself could potentially be the problem is my genre is so restrictive that it's breaking their models though I don't understand why it would work previously and not now. I am specifically trying to produce music that isn't a copy of the modern popular genres.

I have been using SUNO for about 6 months so far.

[frozen dagger]

Sorry you going through that. I hope the team can help you find a solution, or you get it resolved somehow. That sucks...

[polar spade]

Do you have a "trash" file from last week and one from this week you can post just to see how similar the prompting and how different the results.

If you want to DM them, thats fine.

[fiery ruin]

I'll go try to find one when I can and DM you. Thank you.

[frozen dagger]

i wanna hear the specifically not a copy of modern music as well too tho

[fair adder]

the nvda screen reader

[polar spade]

If you feel comfortable posting here, others can review them as well.

[frozen dagger]

i just used nvda and it does read it, although it took me a moment to adjust to it.
Sometimes it is having a delay, and not reading what I am hovering over.
So i played with it for couple minutes. Here is what I noticed. Any of the options that open sub-menus with more options, (like Remix/Edit & Create) are not being read out loud on my mouseover consistently. There are 6 items in that menu list that open a sub-menu with more options, and none of them are being read outloud on mouseover using NVDA. (It reads them sometimes, but not consistently as it does for the other options there.)
It is reading the options in the sub-menu just fine tho. So if I mouse-ver the Remix/Edit the sub-menu opens, revealing the other options and I can mouseover and hear the choices from there.
All of the ones that have a sub-menu are not being spoken outloud, not in any consistent manner like everything else is doing tho.

[fair adder]

i should have precised that i don't use the mouse but only the keyboard. but anyways, like i said, i'll try to send a video recording to support tomorrow

[polar spade]

Sorry. Thought these were quickly available. Post or DM but I may not be further available at this moment.

[turbid latch]

I just wonder if I could've changed prompts or used different ones to get more of the sounds I wanted for my most recent song. It turned out really good of course... Still feel like it might be missing something.... other than a human musician anyway 🤣

[polar spade]

Its hard to say. I'm constantly tweaking even with similar prompts.

[turbid latch]

To be fair, what I made is heavy in the nordic influence and a completely made up language and I needed to specify that r sounds had to be rolled and that I needed an accent lol

literally just sort of put "rolled r sounds" into the prompt and it seemed to work 🤣

[polar spade]

Kewl! At least you didnt have to zoom in and duplicate the R 8 times in the waveform. lol

[turbid latch]

Yeah. It sounds very much like old norse, but in my constructed language

I had to put in several references to nordic stuff in the prompts as well including "viking saga" and "bard song".

[lyric fox]

does anyone experiencing with out of sync words? the Lyrcis start several seconds before the vocals begin. anyone?

[severe narwhal]

hi all, what is the name of the software-tool or website, where you can upload a mp3/wav for analyzing what genre and/or instruments are used? I've seen this somewhere here but I can't find it anymore 🤔 , many thanks!

[fiery ruin]

Here is a perfect example of what I wanted and what was generated bewing very different.

https://suno.com/s/8jlKsSuGNbuX7cYT

Thank you.

Another example.

https://suno.com/s/xM5EOZ4GYPle462t

Another example:

https://suno.com/s/Bh34Gk4IRp8VbRQt

[frozen dagger]

#📚┃suno-school message

[frozen dagger]

like a false start? music is playing, the moment comes when you expect singing to start, and it starts a part of a syllable then cuts off like it was going to start but quickly stopped, then picks up very shortly after.
I've had a few pop up here and there.

[lyric fox]

exactly! did you do anything to fix? is this a known bug?

[frozen dagger]

i was just iterating songs for ideas, wasn't looking to perfect any of it, so i didn't do anything about mine. not sure. you could play the song to where the part you're referencing is happening, share the link with "Start song selected time" and submit a report in the bug channel here on discord.
well, https://discord.com/channels/1069381916492562582/1082775596767117312 or the ios or android bug channel i mean. not here, here.

[copper talon]

I've burned dozens of credits trying to get Jazz instrumentals with NO reverb. The ride cymbals are always too bright and they creep up as the song progresses. Any promopt tips appreciated. I've tried multiple variations of the prompt below

refined dry vintage lounge jazz, electric piano trio, upright bass, brushed drums, laid-back tempo, dry intimate studio feel, dry room, controlled top end, warm and dry mellow tone

[quiet solar]

You could try slipping in something like "close-miked, dry and intimate" (although you're already using dry multiple times, so you might want to be tactical with that). The catch-all "acoustic" can often be helpful for something like this too.

Trying to completely avoid reverb will be hit and miss. Using the prompt above, it was about 50/50 whether Suno went for a dry, reverb-less production or the natural reverb of a close-mic performance on a stage.

Essentially, there may not be a 'magic' prompt for avoiding reverb. All you can do is steer Suno towards things that make it less likely it'll reach for reverb when mimicking production effects. But shifting it from "Does it right 5% of the time" to "Does it right 15% of the time" will still be an improvement for both your credits and sanity.

[polar spade]

@frozen dagger < -- may want to listen as well but they don't surprisingly sound as differentiated as what could normally be expected for Suno to generate. At least in my ears.

The second one, yes, takes a change at the 2 minute mark which indicates a possible and slight Service issue that comes at about 2 minutes sometimes with Suno and Suno sometimes veers a bit. If this has anything to do with a change in the rhythm and sound on this specifically, who knows.

But what Suno also does is take our prompts and, if not a specific genre, makes decisions of what genre they may be a part of, in your examples, and being a fan of New Age, then this to me is part of Native American, New Age, and World. And that encompasses a lot. We can get rhythms and synth drones similar to this from Mike Oldfield and Patrick o'Hearn to name a couple. So, Suno may have gotten a bit more free with the idea of meditation to include the more rhythmic beats found in New Age at the second half of #2

The first iteration also had this rhythm with the flute at 3:44, which that can easily been implicated by Suno on the whole second half of the second iteration thats its a common thing to include somewhere. Very lightly on the first one, and heavily on the second -- the flute was pulsing and not tapping, clicking or a percussion. But the soft-brush sound in the second one certainly could be considered "percussion" so, yea, it didnt exclude it in that sense.

If you haven't played with personas, then that might give you an advantage to make a persona from the first one. You can add the persona to any new generation with the similar prompting and it should help.

Additionally, you can use the first one, paste it twice as an Inspo and run them with your prompts that way as well.

[fiery ruin]

Thank you for the information. I'll give better try and see if it improves my results.

Any idea why the results changed from last week to this week when the prompts did not?

[quiet solar]

This might be a perspective thing but, like Dasvon, my immediate reaction was that in Suno terms all three are actually quite similar. While you know in your head what you're after, Suno can only do it's best to interpret and approximate that based on what you tell it (and what it's learned about songmaking from a lot of music that will include things like that, and things that are very different).

What you're aiming to do in terms of creating a stable pattern for genre/vibe/style is quite ambitious, especially for solo instruments - which is already quite tricky in Suno outside of Studio at the best of times. It might feel that from one week to the next it feels like Suno suddenly "got it" and then didn't so something must have changed, but a big part of that is our brains being pattern-recognising machines that really, REALLY don't like the idea that random behaviour can sometimes give the temporary illusion of consistent results. Especially if we were happy with those results.

On a more practical note, Suno doesn't handle negative or conditional prompts very reliably. You might want to consider taking out things like "no crescendos" (which Suno will often read as "crescendos") and changing things like [Low Register Only] to [Low Register] or [Prominent bass frequencies].

[polar spade]

From Suno's perspective of giving a random spectrum of that style, it probably doesnt thing it did.

If you kept the same Create panel up and ran many output, you're more apt to get similar results from what I've experienced, but if you run at least 30 gens, you'll start getting enough random ones peppered in as well that this feels normal.

But no. No reason "why". Again, I dont hear anything harsh enough to claim its not doing what the prompts are asking .

[frozen dagger]

Nicely said and sage advice, as usual. What I heard I was thinking it sounded pretty normal Suno behavior. The prompts are slightly different but the audio mostly the same. I wasn't sure what we were looking to do here.
Closest thing I got from this was looking for a "pattern" to use, like just looking for a good prompt that created consistent with what you looking for.
So my thoughts went to the Persona straight away before heard anything. Just my preferred method to use audio when I can. (If not a Persona than a track that can influence audio via some other remix (like inspo Dasvon mentioned too) and then use the audio to generate more iterations with similar vibe.

[fiery ruin]

The results are much worse without the negative prompts. I honestly haven't figured out any other way to get close to what I want without them. Suggestions are greatly appreciated.

[quiet solar]

I'm not saying to remove the exclusion prompts from the Advanced Setting field - just to be wary of putting exclusion prompting in the 'positive' prompt fields like Style and Lyrics.

[fiery ruin]

I've tried both to be honest and in both cases it didn't help. I've even tried using different GPT models to see if maybe they could help me refine some of what I was doing. One particular example I got a really bad hiss and I still don't even know where that came from.

[copper talon]

thanks. i've used close mic, i've thrown everything i have at it. The bigger issue is ride cymbal creep for instrumental jazz - ride creeps louder into hallucination as the song progresses..

[quiet solar]

@fiery ruin Okay, best tip I can give you - mentally take a step back. Your individual prompts are influential, but not quite as specifically influential as you might think.

To borrow a phrase from someone else, Suno isn't a word processor for music. You can't really command it to do or not do specific things, especially at a detailed level.

Far more so than working with LLMs like ChatGPT and Gemini, prompting in Suno is much more like giving general suggestions for interpretation rather than specific stage directions. And accepting that sometimes Suno will give something you've said a lot of attention, and at other times a lot less. Lyrics box prompts are particularly prone to that.

If a really bad hiss turns up in one generation, that's just a thing. It happens. It might be a genuine mistake in how it's handled the production effects, or it might be hearing hissing in some of the training data your prompts have steered it to and thinking "Hey, maybe Rose Heart would like some hissing in here too?". Either way, it's generally best to shrug and move on rather than overreact to it.

If a really bad hiss turns up in 8 out of 10 generations then that's a different issue, and is very probably something you've prompted Suno to select for without meaning to. Definitely worth revisiting your prompts and making changes in that case.

[fiery ruin]

@polar spade @quiet solar Thank you both for the tips in trying to help me out in understanding how this thing works.

It's pointing me in the right direction.

[quiet solar]

Yeah, I've had that problem with cymbals specifically as well.

If you're using Studio, I'd say isolating the stem and covering the back half of it can be quite a helpful way to eliminate the issue. If you're on Basic/Pro...I'm afraid I don't have any good answers short of getting Suno to re-perform that section using Extend or covering the whole song on a not-too-high audio influence and crossing your fingers.

If you find a better answer, I'd love to hear about it though.

[fiery ruin]

Are there any particular words or a list of words that I can track down that produced results within the system?

Maybe if I could figure out exactly what SUNO is using for a language, it would help me determine a better route.

[frozen dagger]

@fiery ruin I was going to make this suggestion before. (If you haven't already) spend some time playing with that magic wand in the style prompt.

The simple mode is also another option. Working in the similar way to the magic wand, just speak plainly about what you want your song to be, hit create and let Suno turn that into a song with a prompt. Study the prompts it's making and the outputs for what Suno thinks you're looking for.

[fiery ruin]

I've only run into cymbals one time. Usually I end up with a tapping or clicking sound here and there.

[fiery ruin]

When I first started, I used the style only window. And still go back and forth. I'm going to have to check my site for the magic wand though. I've read about it but never tried it.

[quiet solar]

There isn't a specific list of words that Suno does/doesn't recognise. It boils prompts down to keywords and uses those to look up references in it's training library, but there's no fixed list of terms or specific structure it'll accept like a programming language.

Incidentally, this is why putting "no vocals" in the style box isn't a great idea, since the keyword is likely to end up being "vocals" - the exact opposite of what you want.

There are terms that Suno can recognise more reliably - as a general rule, it's better with musical terms than non-musical ones. So for example, expect it to be more reliable if you ask for a "Soprano" rather than "a high pitched female singer". But both may well work or not work in any specific case.

It's also generally better at broader genre/vibe/style prompts than it is at highly specific direction. It'll get "Thrash Metal" pretty reliably, but is more likely to disregard you asking for a C-E-F-G chord pattern. Being designed as a songmaking tool, it's also heavily influenced by things like genre - if you're prompting to include things uncommon to the genre, or exclude things common to it, your prompts will typically be less reliable influencers.

[potent saffron]

hi any good rock song? with deep meaning?

[fiery ruin]

That makes sense and it does follow a lot of what I have seen in working with the system. How do I get it to adhere to the exclusions more clearly? I do have a pro account.

[quiet solar]

You can't, really. Best advice is to only put exclusions in the "Exclude styles" field under advanced settings and put them in as positives. E.g. if you want to exclude drums, put "drums" into the Exclude styles field rather than "no drums" or "-drums".

Will that actually exclude drums on it's own? Generally not. The majority of things Suno finds when it looks up songs in its training data will include percussion, because it's so common to music across the board. Exclusions aren't given a huge priority over positive prompts in the model (maybe in future...?) so they won't override what it's finding from what else you've asked for.

If excluding drums is a big priority for you in that song, you want to include prompts for what you do want in style that steer it in the direction of music that doesn't include drums. You're already doing a lot of things right there - I'd anticipate things like 'minimalist', 'meditation' and 'soothing' will all be very helpful on top of any specific exclusions.

Personally, I rarely use the exclude field - positive prompting is so much more effective that I can usually get to where I want without reaching for excludes. However, others absolutely swear by it. There is often more than one way to go about getting to the end result you're looking for, and that's what ultimately matters in the end.

[fiery ruin]

I figured a lot of it was the type of music I was trying to produce which is historically accurate and requires a certain formulated styling.

The music I am working to reproduce is more of 1,100 to 1,300 AD, a very specific style and genre but not easily reproducible.

May I DM you?

[quiet solar]

Sure, go right ahead

[rain lance]

I had a hunch putting words into the exclusion field might encourage them more to be in the song, and then I've read other people say the same thing. I usually had better results changing my prompt slightly to try and avoid things I don't want, rather than trying to simply exclude them, it just doesn't work well if at all. I haven't used the exclusion field for months, aside when trying to fix a problem and it never worked, so I just avoid it now.

As for prompting, if you're not getting the sound you want at all, I would try finding a piece of music that sounds just like the sound you want and is well known enough that AI can recognize it by name. Plug that into another AI's text chat and ask it to describe the style of music while leaving out names related to the specific piece. Sometimes that can get your sound spot on. You do have to be careful though, because if suno starts adding hisses to your generations, it might decide that's a good idea and continuously add them to all generations after that that are using the same prompts, cover, personas, etc. I haven't found a solution for this.

Sometimes, less is more too, so not using a complicated AI generated prompt description and just using simplicity might nudge you closer to where you want to be.

[earnest rune]

I just want to say thank to the helper here the other day that told me to delete cache and it would fix remi lyrics.
I had that prompt that popped up today when I got on suno, cleared it, and it fixed the repeating lyrics.

[quiet solar]

It shouldn't do if it's in the exclusion field, but I'm pretty sure it can do if they're in the Style/Lyrics fields.

On the hissing example, it's been confirmed by devs that Suno doesn't have a 'long-term memory' of things you've done previously. If you do a new generation using Create, it won't reference anything you've done before, so it won't remember that it used a hiss once and decide to use a hiss again - though if the prompt brought in a hiss before that's a sign that it could do so again (but so could other prompts, etc.)

Cover/Persona are a bit different, because that will import obvious and subtle elements from the uploaded audio. And those could affect any generation using that input. So if you cover a song or use a persona that's predisposed to hissing in some way, then...yeah, generations with that audio will be more prone to hissing as a result.

[random stag]

How do I add a bio? Or do I have to do it on a computer?

[fathom crypt]

Second option. Maybe on a computer 😂

[random stag]

What ways do you publish your music on a larger platform? I can’t link my Spotify account and can’t figure out how to upload on YouTube music 😵‍💫

[fathom crypt]

You can post your question in #💬┃general-chat maybe someone can answer you. Thank you so much 🙏

[quiet solar]

I'm presuming when you say "Spotify account" you mean your personal one, rather than a Spotify for Artists account?

You can upload music videos to a standard YouTube channel, but if you're talking about uploading to Spotify or to YouTube Music, that's something you have to do through a distributor. That's also how you get access to a Spotify for Artists account and (potentially) get a YouTube Official Artist account if you meet their requirements.

If you're completely unfamiliar with that side of the music business, it's definitely something to Google as a starting point, since it'd be quite a lot to talk through the whole end-to-end process in here. 😉

[quiet solar]

This isn't really the place for self-promotion. Please keep that to the Suno Showcase section. 🙂

[ornate wigeon]

i want song style prompt to generate sad cinematic rap song, do anyone have prompt for sad english song like NF : The search , anyone ?

[stuck junco]

@ornate wigeon Here's a one that turned out some cool results (using v5):

Cinematic anxiety-rap, 120 BPM, dark orchestral trap with tense strings, thunderous drums, distant choir accents, and cold piano stabs, Intense male vocal, tightly controlled, fast and articulate, sounding like a public breakdown held together by discipline, No lush warmth, no playful swagger, no celebratory energy, Build through constant escalation instead of a big catchy chorus, Heavy psychological pressure, intrusive thoughts, fame fatigue, inner conflict, gray skies, empty roads, and operatic tension.

Hope this helps!

[ornate wigeon]

thank you so much dude , really appreciate it , can i ask one more style

[stuck junco]

Sure thing, happy to help!

[ornate wigeon]

thanks dude .
next style i am struck with is :
SUNFLOWER by Post Malone, Swae Lee

these kind of music is hard to achieve

i can do hip hop but getting these standars and style used kind of old days but new style is achieving so hard

[stuck junco]

While I'm looking into this, got a quick question:

Any suggestions on what could we add to assist ewith your creative flow when trying to capture the sound you're looking for? Really trying to better understand what your goto approach would be when trying to bring your song ideas to life.

[ornate wigeon]

it give like rythemic melody or old version hip hop or something else but the song i am inpired (sunflower)is different vibe , its kind of vibe and sounds like we can listen whenever we alone but not sad , its kind of vibe if you can listen the original song ,
i dont know how to explain but if you listen the original song i am talking about you could get idea what i am trying to say

[stuck junco]

Thanks for that feedback, super helpful!

This had some pretty cool results:
Sunlit melodic pop-rap with soft kick, light snare, laid-back bounce, clean guitar, airy ambient layers, and bright warm harmony. Soft melodic male vocal, relaxed and airy, emotionally open, polished and present in the mix. Clear verse/hook structure, smooth and replayable, with a golden-hour warmth that feels light, sweet, and effortless. Heavy hallway reverb vibes.

Hope you have a great day!

[ornate wigeon]

thank you so much

[lone nova]

What do u mean hard to achieve? There must be so many similar songs out there.
Break it down, find similar style and learn how to describe it.
Post Malone is really not that unique, IMO.

is this similar enough? the lyric was a joke, intended for happy hardcore, just for reference.
https://suno.com/song/587b65ad-b655-47c2-b931-bd53ecfd5601
I am sure we can edit the voice to be more similar, but I just want to get the vibe first.

original happy hardcore:
https://suno.com/s/JOvjkkxSjCjyCrF5

[balmy hound]

I have TV series song from 2006 and it was never made full even post-production from some behind the scenes. So I glued together parts of that song to the single audio to give a reference for AI model on Suno website. I used it to recreate entire song (lyrics was available to public for complete version - used only 3rd in TV series) like year ago when v3 model was the best. v3 model was able to capture voice style (almost cloned the voice and background music). But new v5 model cant do this for some reason. It puts completely different voice to the song and accent.

v3 version was good but it had some "robotic" parts so I wanted to re-create it in higher quality.

Is there any way how to tell AI by prompting to capture entire voice style from uploaded file?

[half dragon]

It's possible not by prompting but by creating Persona and then use it -- but having said that, Suno probably will prohibit you to do that since it's a copyright material.

[ornate wigeon]

i asked for specifically the sunflower song style , the songs you attached was not even close , i guess you didnt understand , its okey no worries

[lone nova]

how close do you want to get to the original sunflower song style?
a parody with exact same music but different lyric?
maybe you have find out each instrument in each section of the original, or by using external opensource models to imitate the original but with different melody and lyric.

suno's goal is not to copy music.

[ornate wigeon]

you dont get it again , i never said i want to copy , i said i want same fell , same kind of vibe .if thats what you are saying copy i dont know what to say to you . and for your information i got the correct prompt style , you dont need to worry about it anymore

[young basin]

I'm trying to use the Extend song feature, but I get an error every time I press Get Full Song. All it says is "Error generating clip". I was wondering if I have to do anything on my end or if I need to have a subscription for it to work?

[lyric fox]

Thank you for your suggestion. They sent me back to this channel. lol

[frozen dagger]

That channel is for reporting the bugs. This channel is designed to help users for anything related to Suno (including workarounds for bugs).
My suggestion was to give a link of the song starting at the exmaple where the error occurs in the song. The Suno team members can see it in there and be informed, especially when multiple reports are being made.
@high snow was making a good suggestion to seek advice here. This is a good place to find workaround solutions. Still, if you can, I would share the example in here so others can know what exactly the issue is and offer solid suggestions.
So, here for any suggestions, but I'd give the example of the bug and a short descript of what it is doing in the bug report channel tho.

I can't think of many other than isolating that part where the vocal is, stem split, remove the vocal if possible, or regenerate that section.
i have had the similar error occur, i don't know where that example is, as i was just iterating and moved on like i was saying before.
but in my experience of what you described those would be the first 2 approacdes I would take. but the stem split method requires using Studio, or a DAW to edit the pieces back together. While regenerating the section will cause changes.

It's probably always good practice to seek help here and report the bug tho. If we don't report can go un-noticed, unpatched. Leaving us doing workarounds instead.

[high snow]

When you tagged me, I thought I didn’t smthing wrong. 🤣😮‍💨

[frozen dagger]

not my intent. apologies. 🙂

[lone flame]

Anyone fight Suno on v5 producing high wash, shimmering, excessive high frequency artifacts in their renders?

[polar spade]

Is this consistent on different prompted songs?
and a couple of test songs of just one prompt?
and at different moments of the day /week?

Thats a better way to tell if its purely a loaded Server error or something to do with the song itself. Not every song "works"

Just to compare I just finished two songs since 6pm tonight (fours hours ) and not any fidelity issue. Just Suno skipping about on some of the generations.

[lone flame]

About 90% of renders. From the most intricate prompts to the most basic prompts. I always have to remaster 2-6x or go into Studio to isolate which stem is producing the artifact, and either replace it, EQ it if possible or just delete that stem.

[polar spade]

@lone flame the most basic prompts is "one" and no tags, if thats what you mean. What happens when you test one prompt? Just try "acoustic" to get a possible one to a few instruments and have suno write lyrics of about 1000 characters. Run it for one generation and post. Lets see what happens.

Are all sliders at any time between 15% and 85% and none at 0% or 100%?

[lone flame]

Here's a detailed explaination from Claude on the issue. I've been working with Claude, Grok, ChatGPT trying to find work arounds, but it's something with Suno that's baked in the renders. // Feedback: High-Frequency Wash / Shimmer Issue in Generated Output
A consistent issue across many generated tracks is an overactive high-frequency wash and shimmer layer that appears to be baked into Suno's mastering chain. It manifests as an uncontrolled harmonic excitement in the upper registers — a kind of airy, over-modulated sheen that sits on top of the mix regardless of what the prompt specifies.
For creators producing therapeutic, focus, ambient, or psychoacoustic content this is a significant problem. The high-frequency modulation actively works against the intended effect of the music — it stimulates the nervous system rather than settling it, which is the opposite outcome for anyone building sleep, meditation, deep focus, or relaxation content.
The issue seems to be triggered or amplified by certain prompt language related to space and atmosphere, but even with carefully engineered prompts avoiding those triggers, the mastering layer still applies its own harmonic brightness.
What would help: either a mastering intensity control, a high-frequency presence/air dial, or simply a less aggressive default ceiling on harmonic excitation in the master chain. Giving creators the ability to bias the final mix toward warmth and low-end weight rather than brightness and shimmer would open up a much wider range of functional music use cases on the platform.
This is not a niche request — the lo-fi, ambient, focus, and wellness music space is one of YouTube's largest and most consistent content categories.

[polar spade]

I'm starting outside of Studio, BTW, to start with a clean slate. As for Thursday through Today Studio has had few hiccoughs.

[polar spade]

OK something baked in the renders. Got it.

[lone flame]

I just ran 20 renders, even using 4 clean tracks as inspiration, and I'm having to go into Studio to try to clean up each one. That's after hitting remaster up to 3 or 4 times on some of them.

[polar spade]

all great suggestions.

[lone flame]

There's channels on YouTube that just mass produce tracks to throw into mixes, but they don't clean up their tracks and after about 30s into their mix videos, that high wash is there and I just click right off their video. I try not to let any tracks like that slip into the mixes I put on my YT channel.

[young geode]

Depends on what you mean.

SUNO uses an LLM as an intermediary to deconstruct your prompts and Metatags, for its music ai.

The Style prompts give the AI a broad skeleton framework, while your Metatags act like muscle. So for best results you want to combine both.

Style prompts work best by either targeting an era-genre (1980s Thrash Metal), and specific instrument-models (Guitar: Stratocaster). Doing this helps focus the AIs attention, Otherwise the AI just sees Thrash Metal and Guitar and it pulls from effectively all known Thrash Metal and Guitar. Same goes for tempo and rhythm, 00 BPM rarely works, but if you use the Italian tempo terms like Adagio (66-78 BPM), it gives you a slower song.

Negative prompts work best for targeting genres/eras you don't want to get mixed in. For example, if you try to fuse Metal and Rap together, it'll likely start sounding like Rapcore or Nu-metal, given those are fusions of those two genres, so you'd exclude 1990s Rapcore, 1990s Nu-metal, 2000s Rapcore, 2000s Nu-metal.

Given you seem to be chasing an instrument, go over my songs in this Playlist:
https://suno.com/playlist/f3380023-03fa-4630-a7e0-3f4342dd7f39

You can guide instruments and instrumentals using non-verbal languages like morse code, and tagging specific instruments and their models, with things like named scales and specific named techniques.

[polar spade]

All good intelligent and worthy suggestions. What would you suggest we do here with your question asking if others are getting the same effects? As I've mentioned, I'm one that hasn't had an issue, say the past 4 days, while I've been using Suno and Studio. So thats helpful to know.

I'll sit back then and let others note whether the same is happening with them.

If you'd like to make a suggestion then, the developers are alway reading the suggestion channels. https://discord.com/channels/1069381916492562582/1154587937611190354

[lone flame]

I've noticed that Suno tends to ignore negatives, so I try to leave out anything negative and keep everything positive. It's like it ignores the words no, avoid, none etc

[lone flame]

Yeah I'm curious if anyone else has noticed it. https://youtu.be/J5VRowgI4b8?list=RDJ5VRowgI4b8&t=120 at this timestamp in this video, you can hear it. It's like Suno likes to run open hi-hats, even when you tell it to run closed or tight hi-hats. It does it with a lot of different types of cymbals, especially notice it with Synths even seen it in some Bass stems too.

I've even tried to remove anything in my prompts that would produce too much reverb or cavernous echoes but it's like 8 out 10 renders, it defaults with that high shimmer. Just thought I'd pop in and see if anyone else fights the same issues.

btw, that's not my channel, just a random one I hear the artifact in within their first track.

[young geode]

LLMs don't deal well with negatives to begin with.

If you're trying to fine-tune the sound, then you're better off using Metatags like [Instrument: Drum, Tama Kit, Tight Hi-Hats]

[lone flame]

yeah I've tried that too.

[young geode]

Then check the genres you're targeting.

Genres will often override whatever it is your trying to do, so you might instead look at a similar genre that has the sound you're looking for and target it and it's drums.

[lone flame]

I'll tinker with that, but it's not always a cymbal issue as it happens with the synth a lot, even if I tell Suno to keep the synth low in register.

Appreciate the advice though. Thanks.

[tight prism]

Oh boy, spam.

[honest heath]

<@&1082792303976001547>

[young geode]

If you send me examples, I may be able to help.

Bass, Drums, then Synths are the hardest to control at times, but I have found that if you target specific genres, and then target the specific instrument models, you get a much better result.

[Drums: Ludwig Kit, snare raw cracks, blast beats, Amen Break Beat], [Synth: Roland TR-808 kicks, phonk cowbell accents, rapid fills], [Bass Guitar: Fender Precision Bass '77 thru Ampeg SVT, overdriven driving octave runs, finger picked]

Like that. It takes up a lot of space in the Style prompt, but it works a lot better than just calling out Drums, Tight Hi-Hats

[lone flame]

Unstructured Prompt // Psychill psybient, 76-82 BPM downtempo hypnotic groove, long weighted intro of lush velvety dark pads and dense layered tones slowly building into emotional dark minor-key melodies carried by grand piano with hesitant phrasing and anchored chords, thick warm sliding rolling bass with smooth portamento glides and gentle breathing sidechain, soft grounded kick and laid-back percussion, evolving harmonic textures with body and mass, dense low-mid pressure depth, subtle alien glitch FX with percussive bite, glowing fractal alien vine imagery, swirling layered incense motion, pristine warm mix with deep low-end foundation and smooth balanced frequency response, slow fade ending with sustained harmonic weight and full low-end presence, controlled warm highs with soft natural roll-off

Structured Prompt //Psychill psybient, 76-82 BPM downtempo hypnotic groove
Bass: thick warm sliding rolling bass, smooth portamento glides, gentle breathing sidechain, deep sub presence, gritty saturated mids
Drums: soft grounded kick, laid-back percussion, swung hi-hats, slow half-time feel
Synth: lush velvety dark pads with body and mass, dense layered tones, evolving harmonic textures, emotional dark minor-key melodies, deep low-mid pressure depth
Keyboard: grand piano, hesitant phrasing, anchored chords, slow melodic build from weighted intro
Vocals: subtle chopped vocal textures, warm filtered tone, rhythmic breath motif
Backup Vocals: distant layered hum, dark minor harmony, soft tonal weight
FX: subtle alien glitch with percussive bite, glowing fractal alien vine imagery, swirling layered incense motion
Mix: pristine warm mix, deep low-end foundation, smooth balanced frequency response, controlled warm highs with soft natural roll-off, slow fade ending with sustained harmonic weight

[sweet ivy]

Any suggestions on a musical number, starting with V1-female solo , then V2-a male solo, duet in the 1st chorus, then how could I add 2 more singers in 3rd verse, to have 4 people?

[young geode]

https://suno.com/s/AnQ1t0YKNF5SMTDR

https://suno.com/s/FQy7R4jLa78JfhDP

Close to what you're looking for?

[lone flame]

Not really but I'll tinker with the lyrics/style format you used to get more of my sound. In the second track, at 2:19, you can hear that high wash shimmer that I'm always fighting against. I'll give that format to the AIs I use and adapt it to the music I'm aiming for.

I appreciate it.

https://suno.com/s/Hs33z8tjSX1RU3Ro this is one of tracks I've liked the most from an older prompt.

[young geode]

Well this is the first version, if it's in the ball park.

Im using this as a base to figure out what it should sound like: https://youtu.be/dAA_mdQOXe8

Within the lyric meta tags Evolving Velvety Textures starts at the first chorus and this may be what's causing this issue for you.

[lone flame]

That video is pretty close. https://youtu.be/uGuH3dCSwNs?si=raaK0jjUCJb6iTYS this a recent mix I made with all clean tracks. Style I'm aiming for is like a blend of Nomyn, Marion with the chopped vocals and alien fx of Neelix. Which is basically Psychill or Psybient.

[young geode]

Are these their own genres, or artists?

[lone flame]

Artists. Nomyn and Marion are in the chill ambient future garage lane. Neelix is in the prog psytrance lane. I've been blending elements of each artist to create something which I found out is actually a sub genre called Psybient and Psychill

[young geode]

then your issue may actually be the Psybient and Psychill genres themselves. That specific sound you keep running into might be one of the most used effects in the genre and the AI is running on pattern predictions.

If that effect is often and widely used, then it's going to use it at some point. So you might instead be better off trying to blend the other two, which may not have it, or have it to this degree.

[lone flame]

It's all good. I'll tinker around with your format structure.

I do appreciate the assistance. Those two tracks you linked were much cleaner than some of the renders I've been getting. So even with minimal artifacts, I should be able to isolate and EQ it down on the high frequencies

[young geode]

Tell me how it plays out.

What I gave you was a basic format for the metatags, if it works better (I find a massive difference), I can give you the more detailed tagging method.

[lone flame]

Thanks!

[quiet solar]

It's not easily done.

Pushing Suno to go beyond two distinct vocalists is incredibly tough. You can have one of those 'vocalists' mimic a group of backing singers or choir, but I've never seen anyone successfully split them out into individual voices via prompting.

A three vocalist song is something I've only ever been able to achieve via editing.

[young geode]

It's possible to do dual/multiple vocalists consistently, however it's more about the lyric structure than it is the style prompting.

Example: https://suno.com/s/ZKubMpl852x04Ctd

A combination of Hard changes in tempo, rhyming scheme, enunciation, mood and structural breaks such as "Quotations", (Curves) and/or /Stars have a significantly higher chance of triggering different vocals: especially with genres that regularly have multiple vocalists.

@sweet ivy

[quiet solar]

That depends what you mean by consistently. Getting multiple vocalists to consistently appear in a song is very doable, though having more than two distinct voices makes it a much harder exercise. And then trying to direct them to sing particular parts reliably is a whole other ballgame.

For example, in the song you linked Suno quite happily ignores your direction for a choir throughout, brings the female in on Verse 3 where you call for a solo male singer, and brings the male singer in for the female parts in the second chorus. All of which is absolutely typical Suno duet behaviour, as the level of control we can exert is limited.

I totally agree with you that lyric markers and structure have an influence too. Ultimately, if anyone is attempting something purely through prompting then I'd personally suggest blending style prompts, lyric tags and markers to steer Suno as much as is possible to where singers should switch over or where lines should be handled differently. And still expect regular errors.

The showcase overlapping three vocalist song I point to is below, which did all of the above along the way while building it, though lyrics were cleaned up for the final version. But as far as I'm aware, something like this is only really possible through a combination of prompting and editing:

https://suno.com/song/2c610c73-d5c6-4484-9f58-e3699af2590e?sh=VuzFf8KyT2nzpjou&time=55

[young geode]

By consistently, I mean you can introduce two singers without issue. Controlling them to a high fidelity is different, but getting them in there and consistently singing is fairly easy.

SUNO will ignore metatags, kind of, but not wholesale.

What SUNO will do, often, is match a specific kind of speech/flow pattern with a specific vocalist, then it'll repeat that vocalist for those kind of sections/structures.

That's why it repeats the Male Vocalist and Choir for the chorus without issue.

For that Gen, it did follow the prompts, on previous iterations, so what it did was flip them for this, however, the overall sound of the song in the those iterations wasn't as hard as this one, so I chose to take it because it didn't matter to me which vocalist actually sung what.

And later in the song you see it does follow the instructions for Verse 4, without issue and produced a completely different female vocalist, but returned to the original Male Vocalist for Verse 5, because the song returned to something closer to his previous sections.

In later songs, and SUNO versions (this was 3.0), it's become far easier to control this kind of hallucination, by doing as I said above: especially if you explicitly use only one kind of meter line for one singer, and another completely different meter line for another.

[daring wharf]

Not just you. And I'm subbed, so it's not that. Afraid I've no answer for it, though.

[worthy kraken]

This happens to anyone?
100% of the time when generating ending from the specific song

[viscid ravine]

I want a possible 60 minute/1 Hour soundtrack in SUNO AI, how do i make one @polar spade

[viscid ravine]

yeah, i mean in Suno itself

any tips or?

[sweet ivy]

@young geode That's incredible. It's not a Broadway style like I'm trying for, but it gives me some guidance on using vocal prompts. I assume the word "choir" would be preferred over "ensemble".

[young geode]

Haven't tested it, but Choir works.

Note: this is an older way of prompting from 3.0.

[polar spade]

Try again. It may be functioning better now. Had some good reports this morning.

[young basin]

It actually does work now! Thank you for letting me know :)

[sweet ivy]

@quiet solar That's a wild song. I clealrly hear the male and female singers. See a detailed style prompt, but I don't see any singer tags in the lyric [Female, aggressive] etc ?

@polar spade In exporting a song, it's it better to download the wave file and them run it through Audacity, that has the LAME for converting to MP3 or to download as a MP3 from Suno. Which option will more likely preserve the quality of the wave file ?

[tight prism]

Well, to preserve the quality of a wave file you download the wave file. Downloading the mp3 compresses the data.

[sweet ivy]

Goodness, I downloaded the wave and is was at a significantly lower volume and was flat sounding ?

It was a Remaster, could that have anything to do with it, though it sounded great when I played it from my library

[polar spade]

IN the past, users have noted the mp3 having some differences to other compression programs.. That being vague, it seems reasonable, to download the WAV version if the produced MP3 seems compatible with other play sources. LAME is reliable and considered flexible for its conversions offered.

Find a quiet time to download WAV files so between 12pm and 7pm EST is probably the worse.

[sweet ivy]

@polar spade Maybe I will do what I did with my last song. Extract the stems, copy and paste into my DAW. Export as a wave file. then drop into Audacity for the MP3 conversion

[polar spade]

When doing that, also export the whole song as a file from within Studio and download that one also to see what difference, if any, it makes in volume and or quality to the one in the Library you have thats straight from the remaster.

[sweet ivy]

@polar spade Okay, pull the Remaster from the Library into Studio , then download the wave file ?

[quiet solar]

The song was taken out of Suno into a DAW for editing and mastering, and for including things like the channel shifting effect. The style prompt you see is Suno's best attempt to describe it once it was uploaded back into Suno again - it's not actually what was used to create the song.

There were singer tags when it was being created, and various markers to help steer Suno as to how to perform it, but the song was also put together through editing together pieces from about 11 different generations (see image):

[polar spade]

I was suggesting to export it first BACK to the Library. But you can download a second time straight from Studio to compare the three files.

1• Remastered. D/L WAV from Library.
2• Drag to Studio. Export back to Library. D/L from Library.
3• Drag to Stduio. D/L directly from Studio.

Ill make a note to do the same.

[quiet solar]

@sweet ivy If it helps, the workflow behind it is described here: https://discord.com/channels/1069381916492562582/1466598465969590465

That workflow thread also has a simpler duet created through prompting and covering (see below), which has the original style prompts and lyric markers you can use as a reference. Note that Suno gets it about 90% right, but there are still a couple of mistakes around who sings what compared to what is asked for - the only way around that is either a huge amount of patience, or a bit of time in editing:

https://suno.com/s/gcoaIGNUXaPxfZSm

[polar spade]

What's ytour DAW you use? Someone may help with which tools avaialable in doing an inversion to null test the three versions for any differences.

[sweet ivy]

@quiet solar Wow, that's a lot of work in your DAW. The track you sent clearly worked, Suno followed your prompts for when the female singer sang and when the male singer sang. Thwe female was a lost a bit in the duet sections, the male voice was so strong

@polar spade I use Cakewalk/Sonor

[quiet solar]

Yeah, Grounded is one of the closest examples I've seen of Suno performing what was prompted for without using editing. But that still took using a few covers as an audio guide for who sings what to give it a starting point to make further iterations from - and exactly as you say, it's by no means perfect.

[sweet ivy]

@polar spade I sent the song from the library to Studio. Sounded just the same. Saw the Export button, clicked full song and was expecting a download window, where I could send it to my computer. All it did was send it back into my library. Did I miss something, could I download it outside of Suno from Studio ?

[polar spade]

Right. It just sends to Library. Then download from Libary as usual.
Compare the two. See which you like, if any difference. At least compare the volumes which you mentioned seemed lower on downloading.

[sweet ivy]

@polar spade Not much difference. I'm going to forget the Remaster and download the original gen, see if that preserves the volume and brightness. It's insane that I tried twice to download the Remaster, originally from the library and again from the library using the Remaster , Studio sent to the library. I noticed the file of the download looked low for a wave file - 39 MB ?

@ This is crazy. I downloaded from the original gen and the same thing. I played the download for 20 seconds and then the original in the library. Different as night and day. Makes no sense, the download as a wave such be close to the original???

[tight prism]

Is the song 4 minutes long?

A wave file is typically 10MB per minute

[polar spade]

And the WAV downloaded directly from Studio? Same volume?

[sweet ivy]

No, 3:36

[tight prism]

OK so 39MB is about right

[sweet ivy]

Yeah, even the original from the library, was clearly lower volume and flat

[tight prism]

<@&1372996242325508176> what format does Suno work with on the backend?

[polar spade]

Directly from Studio... as in the "download WAV " within Studio?

[polar spade]

That I dont know.

[tight prism]

Because downloading as a WAV still might result in conversion/compression though I haven't really seen the lowness difference.

[polar spade]

Yep. Im trying to see if Studio has its own conversion.

[sweet ivy]

@polar spade Download wave file from the origiinal in my library. Maybe I will need to extract stems, import into the DAW and export from there

[polar spade]

Ill do my own testing when I get in front of my computer.

[polar spade]

Yep. You mention that one. I assume that's version #4. 😉

[sweet ivy]

@ No it's 5.0

[simple briar]

Yeah, I’m not fully sure if Studio has its own conversion built in. You might need to export and convert it separately depending on what you’re trying to do.

[polar spade]

Oops.... my bad... I meant four patterns of download to compare.

[sweet ivy]

No three

Over an hour ago, I played the Remaster, sounded great, thought I would simply download and upload into Audacity to createa MP3, still trying to get a true rendition of the Remaster, or even the original, at this point . LOL

[tight prism]

What happens when you download an mp3?

[simple briar]

Ohh got you 😄 yeah that makes more sense. Comparing the different download patterns is actually smart.

[tight prism]

If the only thing you're doing is pulling it into Audacity to convert to mp3, just download the mp3.

[polar spade]

So yea, do the downloading of stems first. We always should consider some things/ songs/projects may work better with Suno, and some outside in a DAW, so really no expectations, we just make a few tests to determine which is better and just flow in that direction for that project. .

But no, not sure why your volume/amplitude is less from a Library download of a WAV file compared to the playback volume heard within Suno.

[simple briar]

Yeah that’s true 😄 makes sense to just download the mp3 directly instead of converting.

[simple briar]

Yeah that makes sense, testing both and seeing what works best for each project is the smart way to go.
And yeah I’ve noticed that too sometimes the WAV from the library can sound quieter than playback. Might be normalization or how Suno processes it internally.

[sweet ivy]

@tight prism I believe download as an MP3 directly from SUNO will even lose more edge, that's why most advise to download as a wave file, since it is a wave file in Suno

[simple briar]

Yeah exactly, MP3 will compress it more and you lose some quality. WAV keeps more detail since it’s uncompressed, so it’s better for editing and mixing.

[tight prism]

No, it was advised if you were going to pull it into Audacity or another DAW to use wav.

If all you want is the mp3, just download the mp3.

Plus, we're now troubleshooting.

Does the mp3 sound fine when the wav does not?

[simple briar]

Ohh yeah, that makes sense. WAV for editing in a DAW, and MP3 if you just need the final file.

[polar spade]

Have ypu doubled the track in your DAW to get a richer volume? That may be all you need to cover you delimma.

[tight prism]

We haven't heard either version yet.

[simple briar]

Sometimes an MP3 can sound okay even if the WAV doesn’t, but usually WAV preserves more quality. It depends on what’s causing the issue in the first place.

[tight prism]

@sweet ivy What are you playing the wav's in locally?

[polar spade]

Yea. Based on their perception. Im staying in the K.I.S.S. lane here.

[simple briar]

yea

[tight prism]

Like was the browser player turned up to 10 and the local play at a 6?

[sweet ivy]

@polar spade In both of the last two projects I did, I didn't even try and export as a wave from the library. I extracted stems, downloaded into DAW, everything was exactly like when I played it from my library

[lone flame]

Using your format structure, I've gotten a well nailed down formula that's pretty much eliminated about 80% of the high wash shimmering. I'll keep tweaking, but at least now I'm not burning through credits with low quality results. Thanks again.

[sweet ivy]

@tight prism Volume level on computer was set at 100, I checked the sound levels , in Sound Settings

[tight prism]

What program you are you playing the wav files in locally?

PC volume can be overriden by application volume.

Have the PC volume cranked to 10 doesn't matter if the app has it muted.

[polar spade]

Right. Cool. Thats the way you should keep doing it until the reason why the other way isnt working out. We went through some minor tests. So, mention it again layer if nothing improves and see what anyone else can work out. Meanwhile ill test as well when available.

Ill let this cotinue with @tight prism as ive got to run. Best of luck.

[sweet ivy]

@tight prism Oh, I'm playing them in Windows Media, but that setting is also at 100

[tight prism]

OK didn't you say you had Audacity?

[sweet ivy]

@polar spade Thanks, great input as usual

[tight prism]

Windows Media is kind of a low end generic player in the grand scheme.

[young geode]

Good-good.

If you need any more help, just hit me up. Most of what I do is tinker and develop prompt, lyric and Metatags for this very reason.

[polar spade]

We can, if need needed, change the overall volume in studio befor exporting, just mentioning, in case its not known. Bottom right volume slider (may need to hide the track info panel. OK. Now im gone.

[sweet ivy]

@tight prism That makes sense, I don't need to do any editing. I could download as a wave, send it into Audacity, hopefully it will sound very close to how it sounded in the library and I can convert to Mp3 there, which was the plan

[tight prism]

If you just need the mp3. Download the mp3 version

@sweet ivy you are adding extra steps to get a version already available to you

[sweet ivy]

@tight prism I know I can download the remaster from my library as a MP3, and I can try that, but I'm sure there will be even more compression. The LAME system in Audacity works great because I can use variable bit rate, adjust the quality up to 220 - 270 kbps

[tight prism]

Does the mp3 version sound right? You still haven't answered this

Does the wav version sound right in Audacity?

[sweet ivy]

@tight prism Actually the MP3 download sounded better than the wave download, which is odd. I see it was ab 4 MB. With the settings I outlines in Audacity it end will end up @ 6- 7 MB and have a little more brightness. Old studio ears, maybe I'm being too picky, the Suno MP3 download was okay

[tight prism]

@sweet ivy Did you play the wav file in Audacity? Yes or no?

A wav isn't an audio codec like mp3. wav is a container/wrapper. A wav file can contain an mp3 file.

[sweet ivy]

@tight prism Not yet, did the MP3 download from Suno first. The download, whether wave or MP3 seems to go straight into Media player format

[tight prism]

mp3 = audio codec, wav = container that holds an audio codec.

Windows Media needs to use the meta data in the wav container to determine how to play back the audio.

bitrate, channel, etc

Windows Media could not be handling that meta data properly.

Because it's a free, built-in media player. More advanced players may handle it differently.

[sweet ivy]

@tight prism I did play the downloaded wave in Audacity and it sounded much better.. How could I replace my Windows Media player with a better player, that would also be free ?

[tight prism]

You go looking for one that you like and does what you want

[sweet ivy]

@Okay, thanks for the help. Gotta run for now. I can use what's in Audacity, the remaster wave file, then convert using the VBR and quality at 220 - 270 kbps for the MP3 conversion

[high snow]

lol, all good 😊

[worthy kraken]

I need help from the people experienced in the Song Editor please.
I have really weird bug that i can't explain and don't know how to fix

[frozen dagger]

"can't explain" is an issue for us to know how to help. canu share screenshots or link to song where the error occurs? something to show us?

[worthy kraken]

Not sure if the screenshots will help.

The problem is that the lyrics shown in the Song Editor are wrong. Like, the song starts with the pre-chours for some reason.

It does not affect song itself, but only until i try to replace section. Then it uses the wrong lyrics for replacement.

The weird part is that if i use the "change lyrics" section, it becomes even worse for some reason - first, it replaces much bigger section than selected, second, it completly ignores the lyrics written in the replacement section and still continues with wrong lyrics in the song

I tried to edit the displayed lyrics in the main second of the Suno, but the lyrics inside the Song Editor are still messed up

Is there a way to either fix the lyrics, or somehow replace a section without triggering the bug?

[frozen dagger]

the replacing sections part of the problem could be the mode (smart, classic or fixed).

there is a match lyrics to audio. in the new editor mine is greyed out, but if i go to the legacy editor the option is available there.

you may need to save the changes you made to lyrics first and then bring the new save back into editor with correct lyrics first?

[worthy kraken]

Yes, thank you a lot!!! Somehow i tried different function to that, the one in the lyrics editor, and it did not worked.

But how do i save the song afterwards? If i switch back to the new editor, the lyrics becomes messed up again

[frozen dagger]

just Save as new song. then bring the new song in.

[worthy kraken]

I cant find the option to save in the old editor...

[frozen dagger]

interesting... the button has been removed. exploring.

[worthy kraken]

But i think i can just try to generate replacements in the old one.

Then i will be able to save the new version.

[worthy kraken]

The UI shenanigans

[frozen dagger]

yes sir. this is very confusing. I thought if i tried to do a crop (of the full song) then it would lead me to saving my new cropped version. i don't see anyway other than taking it back into the new editor to save the changes. was not like that before. so if taking it back into the new editor undoes what you're doing in legacy editor, then that is very frustrating indeed.

[polar spade]

The edits in editors should automatically appear in the Library. Check the Library for the "saved" file.

[frozen dagger]

i am. refreshing "My Workspace" unchecked filters, looked in the folder i was working in before. nada. Glad to see you here. You're way more familiar with editor than I. Looks diff to me now too.

i thnk i foudn them in a diff folder

[worthy kraken]

Seems like lyrics edit does not count as edit?

[polar spade]

We're making an edit, such as "crop" and we cant find the cropped file when done? And not simply in the Library at the top as s new file? Is that it?\

[frozen dagger]

<@&1372996242325508176> originally we are helping @worthy kraken with their issue.
#📚┃suno-school message

[worthy kraken]

I also facing another problem... now all my attempts to edit greatly misalign with the rest of the song. Like, the system does not tries to fit them it, it is just kinda cuts the song and shoves the part there. But the moment of cut is clearly audible either by music, or by the loudness level. Anything can be done with that?

[frozen dagger]

I will figure out why edit saved to the diff folder later. lol.

[frozen dagger]

that is a blending issue. it does a better job with transistions in Studio doing the same thing, but even then some extra steps are needed for a smoother transistion.

[polar spade]

Ive abandoned lyric editing as a thing.. haha... There seems to always be a bit about Suno pulling from previous generations, literally the previous iterations and combining lyrics from older edits or the original. Its done that since inception. It's now just habit to always highlight and copy each time we have any new edit to lyrics, before clicking any save or confirming any edit. Thats' my suggestions. Have google docs, an open email draft page or Notepad.. anything and keep a constant vigil of copying the latest edit before committing to anything. If anything is off in the saved lyrics, then just paste the most recent copy back in. its the equivalent of smacking the TV on the side to get better reception. You just have to do it.

[sweet ivy]

@polar spade Circling back. Think the issue was more about Windows Media player move than anything else. Dropped the wave file in Audacity, created a MP3 from there. When played through Media player is was a little lower But when I attached it to an email and played in Preview, before sending, it sounded very close to the wave.

[polar spade]

Excellent. 🤘 Next suggestion was to make sure all other browsers and apps were closed. From there, find a different playback module. But you got it then. Well done.

[sweet ivy]

@worthy kraken I'm certainly not as experienced as @polar spade or @frozen dagger , but I has an issue replacing lyrics in 5.0. Someone online suggested using 4.5. I did and it worked fine, but your situation sounds more complicates tha replacing a few words

[polar spade]

I'm happy to learn or be reminded of what works. So. thanks for letting me know what came of it. I'm still looking to test a few things on downloads, and they

they've recently said they un-kinked studio this weekend. So good time to test more.

[worthy kraken]

Damnation... spend half of the credits to make a song, hoping that i will fix the intro later.
Well, that was an expensive lesson.

[frozen dagger]

it may not have had enough space in the edit to replace. (why the mode is important) so, with that try to add a blank section in the edit to give it more room can sometimes help too.

i'm usually doing a cover of the end result to remove those bad transitions. my goal is normally to get it close enough for a good cover. (not for everyone, wont keep that "perfect" copy we fell in love with)

[worthy kraken]

I tried the mode without fixed duration, same issue.
And as a bonus it hallucinates a minute-long intro >_<

How do you add blank section in the legacy editor?

[worthy kraken]

Sadly even remasters remove a lot of nuance. Covers are mostly terrible( Especially after you fine-tune every section

[frozen dagger]

click the little plus icon betwen the section. or creat a split then add the blank section at your split.

[worthy kraken]

Can't see the plus. Or the split. Are they in the old editor?

[high snow]

That doesn’t seem to work for me. Like it will add it but it completely ignores my lyrics and doesn’t sound like the rest of the song 😑

[frozen dagger]

trying to test. and the editor is freezing on me. this why i avoid editor. =\ well, i was able to create the blank section. typed in some lyrics, it did completely ignore. it used the style prompt to create the music, not the song track. but i kinda expected that, i was trying to select the new section along with part of the other song to try and recreate with the music. which after i did that i thought i could have just duplicated the section so it would already sound like the track, and i could adjust the duplicated section's size and just recreate that part; hoping that would keep the musical style already there.
It's been a while since I been in editor.

if it ignores the lyrics, and the other issues tho... would be moot point.

[worthy kraken]

Great. Now after changing one section entire song became a single section)

Now it does not even want to match lyrycs with audio...

I suppose i am the only one here who encounters that much bugs per day >_<

[quiet solar]

I've found that if the Editor/Legacy Editor starts breaking on a song then you can expect it to keep on breaking in curious and innovative ways on that song going forwards.

The issue you had with harsh cuts and difficulties in doing subtle corrections using the Editor is that reason that back in pre-Studio days I stopped using it, and switched to using Extend or Cover to create samples for fixing and patching them in using Audacity, where it's easier to manually blend the edits together seamlessly.

[flat needle]

I just generated multiple songs with this style description: "Cinematic, ritualistic atmosphere with a palette of tension into grandeur into release, Texture blends layered guitars, march-like powerful, intense, dynamic drums; distant, echoing vocals—evoking a mythic, processional feel, 120 BPM, steady 4/4, Drums drive momentum: punchy kick, wide-reverb snare on 2/4, tight hats opening in choruses, Centered in D minor, Main progression: Dm–Bb–F–C (notes: D, E, F, G, A, Bb, C), Guitars use tight power chords and ambient layers emphasizing D (root), F (minor color), C (epic feel), with sparse leads rising F→G→A, Focus is space, not virtuosity, Subtle pads reinforce D, Bb, F, adding film-score weight, Vocal sings: A, C, D, F, Verses: A, C, D, restrained and intimate; choruses rise (D, F, G, A), open and choral, Arc: contained into expansive into commanding"

My problem is, that each generation starts with vocal singing this line: "Vocal sings: A, C, D, F, Verses: A, C, D, restrained and intimate; choruses rise (D, F, G, A), open and choral, Arc: contained into expansive into commanding" instead of singing my Lyrics which there are plenty. It even generated barely 2 minute long song. Every other generation is at least 4:40 long. Any idea why does Suno not want to follow my instructions?

[neat nexus]

Cannot reproduce. Don't say that this is doing the notes but it's definitely not repeating the style prompt.

https://suno.com/s/uSZxMuS8xZZg1qPd

Trying again with some metatags:
https://suno.com/s/vViXKelS2YvSmQ6P

An attempt with completely different lyrics. I don't think it followed the style prompt that well necessarily but it also didn't sing the style....

https://suno.com/s/9bOGQHUCI9vf9eAN

All told I ran it six times and did not encouter it singing the style.

[quiet solar]

Sounds like the core chords are Dm-Gm and possibly a Bbm or two. The other chord instructions are being disregarded.

@flat needle While Suno does pick up key signature, I'll be honest and say I've never seen it take on board requests for specific chords or progressions. If it does, it's certainly not reliable. You could strike out that entire section and I don't think it would have much impact on the song itself.

[flat needle]

I asked another creator if Suno takes into account the chords written, and they said yes, so I tried to do the same.

[neat nexus]

It might work better if they can create something with those notes and then cover that to introduce and strengthen the sound they're looking for. Maybe if they find some online keyboard or something that can save an MP3 or something?

[quiet solar]

In which case I'd love to see songs where it reliably does it, because I've only ever seen songs where it doesn't (irrespective of what the creator claims).

[neat nexus]

But it's odd to me that you were getting the stye prompt repeated.

[brittle linden]

I've a old kids song I'm trying to cover. first verse features guitar, second keyboard etc. I've tried putting tags or putting things in the style, but it just mixes them all up and its a jumble of instruments. Any idea on what to try (Im on pro BTW)

[quiet solar]

Yeah, Suno singing from the style prompt is very, very unusual behaviour.

[flat needle]

I can play the chords on guitar, to make an audio reference, I was just skipping this to save credits if Suno could manage it for me.

[neat nexus]

Is the song public domain? (Suno doesn't allow even public domain covers but at least that covers you legally.)

[quiet solar]

That's 100% the way to go. Or hum it, if you prefer.

[brittle linden]

Its one of mine

[neat nexus]

If it's your own personal song you might also get luck by introducing an audio sample.

I wonder if mashups would give benefits for you since you're basically switching?

[brittle linden]

I have uploaded the track, so it should have a frame of reference

[echo halo]

I’m having an issue with a song I generated. One of the tracks disappeared, but before it vanished, it wouldn’t play and stayed at 0:00 the entire time.
Now I’m experiencing another problem. When I try to remix a song that was already generated, I receive a “Copyright Error.” The lyrics are completely original and were written by me. The original version of the song worked without any issues, but the remix will not process and instead gives this error.
Could someone please explain why this is happening and help resolve both issues?

[flat needle]

I discovered, that if you try to cover an audio uploaded, it is better to slide the audio influence to 100%, otherwise it will shift or even create its own melodies

[brittle linden]

OK, I did set it high, but didn't try 100%

[neat nexus]

Argh...be really careful with sliders. They can help sometimes but they can also cause other issues.

Many of the people having bad results all the time are using sliders and using them at extreme settings.

[quiet solar]

Yeah, the tricky part will be persuading Suno that it wants to repeat your exact instrumentation rather than pick up the instruments and do it's own thing. Even on high audio influence, Suno will try to make a complete song based on genre influences, not just what it hears in the audio feed.

I'd also second Hepcat's point. Unless you are 100% sure what you're doing, keep the sliders out of the red zones.

[brittle linden]

I don't mind spending a few credits to give it a go. I've been trying for hours and all I've got is useless. Seems unable to follow instructions

[neat nexus]

Hmm...what does it give you as a style on the upload? Try using that?

[quiet solar]

@brittle linden If you explain what you're trying to do, and share the prompt you're using, we may be able to help.

Worth noting that Suno isn't quite as...'instructable'...as other AIs. You can't really dictate to it as if it's a session musician, you impact it by steering what it should look up as references in it's training library when creating your song.

[flat needle]

Usually I avoid red zones, but for audio uploads, I prefer 100% because how often Suno totally changed what I played on guitar. Even with 100% it still did a small change, but going from 85 to 100 was a big difference.

[neat nexus]

Nothing wrong with using the sliders as long as you recognize that they can cause issues, especially at extreme settings.

I make use of them myself sometimes, in just this type of situation. But some people use them at extreme settings and then blame the tool for doing what it's been instructed to do.

My hammer makes an excellent tool for pounding nails. It's not so great at cutting wood though....

[brittle linden]

for what its worth... "Rock song featuring a prominent electric guitar riff, a steady 4/4 drum beat at 110 BPM, The arrangement is layered progressively, first verse is guitar, second verse is keyboard, third verse is mandolin, fourth verse is trombone, Final verse is drums only, It must use these instruments for each of these verses and not mix them up, The final verse is all together with each instrument being one line of the verse, The vocal is a clear, enthusiastic male tenor with a friendly and instructional tone, The harmonic structure follows a simple I-IV-V progression in a major key, creating a joyful and accessible atmosphere, Production is clean and polished, with light reverb on the vocals and a balanced mix that highlights each instrument as it is introduced, The song concludes with a full ensemble reprise and a final drum fill"

[neat nexus]

That might be dificult to pull off.

Definitely let us know if you succeed.

[brittle linden]

I might try studio and see if I can break it down a bit

[quiet solar]

The first issue sounds like the track just broke during creation. It happens from time to time - Suno typically deletes it and refunds your credits.

If you're hitting a copyright error, that could be a result of the lyrics or music. Though it's unlikely to be your lyrics, if they passed fine first time.

I can't say what is causing the problem, but on the musical side it's uncommon but certainly possible that when Suno generates your song, it'll contain a musical phrase which is highly similar to a copyrighted piece of music. If Suno later brings that in as an audio reference (e.g. Cover, Mashup, Inspo, etc.) that might trigger the copyright detection check and block it from being used. It's possible you're running into something like that.

[toxic bear]

Can somebody Tell me where the get whole Song on the Android App is?

[quiet solar]

Yeah, that distinctive split of instruments is going to be really tricky.

If you had Studio, that would 100% give you some extra options. You could generate the whole thing using a single backing instrument, then cover that instrumental stem for the other verses and turn it into a different instrument. And for the drums only verse, just delete all the other instruments. The final verse might be a bit finnicky, but it could definitely be done.

[half dragon]

It's no where to be found on Android app. You might have to use web browser for this functionality.

[toxic bear]

Is this a Bug or on purpose?

[half dragon]

Honestly I've never use android app to Extend; so I couldn't tell if it was there before or not.
Maybe it's not supported and not available.

[quiet solar]

Extend doesn't exist as an option on the app version that I can see, so I'm guessing it's on purpose.

[toxic bear]

Thank you for the information😊

[frozen dagger]

You just came across that a feature that is lacking. You're right. The android app does allow to extend, it is inside the remix option, but there doesn't seem to be a get full song from that. You can use a mobile browser to get the full song. The app is a work in progress as far as I can tell.

[quiet solar]

Interesting. Absolutely no sign of Extend on the iOS version at all.

[toxic bear]

Yeah and there is my next problem 🤣 the moment i log in in the mobile Browser it sends me back to the app🤣 so tomorrow on my pc🫩

[frozen dagger]

had to click remix. i went ahead and did a remix in the app just to check what happens.

[quiet solar]

My only Remix options on iOS are 'reuse styles' or 'cover'. Neither leads to anywhere you could switch it up to an Extend. xD

[frozen dagger]

it's interesting to me that the iOS would miss it. they seemed to favor that os over android. iOS had the Suno video to song feature. I think that was removed since.

[toxic bear]

On Android you can extend but with 30 seconds without the get the whole song function it is completly useless🤣

[quiet solar]

Yeah, that sounds like an oversight. 😵‍💫

[sweet ivy]

@quiet solar Back working on the duet with adding an ensemble in the final chorus. Thought I would check Suno support and getting one odd instruction - start with singer(s) in bracket [female singer]. then under that Verse 1, but no brackets around 'Verse 1". Does that sound right ?

[quiet solar]

It's not how I'd personally structure it, but there isn't only one way to do things in Suno.

I'd suggest doing something like:

[Verse]
[male tenor] Your lyrics here
[female contralto] Your lyrics here
[both] Your lyrics here

or if the whole verse is sung by a particular singer, then

[Verse - male tenor]
Your lyrics here

However, it's probably a good idea to keep it consistent with the approach you've been using up to this point if you've already been iterating on the song.

UItimately, unless you're really starved for credits, there's no harm in experimenting with different approaches. Just bear in mind that none of them are perfectly reliable where duets are concerned, and you're trying to specify exactly which singer(s) do or don't sing at any given time. Choruses are more likely to bring in harmonies and ensembles naturally, but it won't do any harm to directly prompt for them in lyric tags and the style prompt to reinforce it.

[sweet ivy]

@quiet solar Thanks I will follow the odd instruction from Suno support & see what I get

[male voice]
Verse 1
Lyrics here...

[male voice][female voice]
Chorus
Lyrics here... (add a note: "singing in harmony" directly under 'Chorus', outside the brackets, before lyrics)

Final Chorus
The entire ensemble sings in powerful harmony
Lyrics here...

[quiet solar]

Interesting. I would have said that stands a good-to-fair chance of Suno singing your prompts.

While Suno can pick up contextual clues in the lyrics, in my experience a good general rule is that anything you don't want it to sing should be in [ ].

[sweet ivy]

I think support is way off. I have gotten bad advice from them before. I'll follow yopur guife from the dong you sent Grounded, you were able to separate the female & male singer successfully

[sweet ivy]

I don't get this, but this is straight from Suno support "To clarify: the official Suno formatting is to ONLY use [male voice] and [female voice] in brackets at the start of each section to assign vocalists. Section headers like 'Verse 1', 'Chorus', or 'Final Chorus' should NEVER be inside brackets, they go on their own line above the lyrics and Suno uses them for organization, not as lyrics to be sung."

[sleek stump]

[quiet solar]

Well, I won't knock Suno Support - they have to deal with a huge variety of questions and challenges from tech issues to 'how-to', and I genuinely don't know to what extent they're expected to be 'super users' of Suno in terms of prompting.

What I can say is that Suno is pretty good contextually with things like "Verse" and "Chorus" and in the format they gave those lines don't get sung. And having done a short test, the format of putting [male singer] or [female singer] before a line seems to work just as well as the approach I usually use, at least for a genre where multiple singers are common. Which is pretty much what I'd have expected, since the only difference is what line they're on.

However, almost all of those tests sang the "singing in harmony" note as lyrics. And I've never known Suno cause problems with [Verse] or [Chorus]. It's used very, very widely by a lot of people, including a few of the Suno devs. So there might have been a bit of overconfidence in the language they used.

[half dragon]

Did you email support asking and they replied with the instructions? I've never heard that before. Giving specific prompts nstructions.

[sweet ivy]

@half dragon Viking yes, since I knew it would be more complex than my previous songs, single singer, I thought I would ask.

[half dragon]

Usuallly they'd suggest to ask prompts related questions to the community. That's why I found strange.

[sweet ivy]

@quiet solar I used your format on the lyric sheet - [Verse 1 -female voice intimate, clear] In the second version they almost followed my guide, except it started with a male, probably because I'm a guy & was singing on the vocal track. Solid Broadway arrangement, full orchestra, big ensemble. I put the singing instructions in the Style box - "Duet, male and female voice; 1st chorus: singing in harmony, duet vocals; 2nd chorus: ensemble vocals, small group harmonies". was going along great until they sung a different melody of the last line of the chorus

Had the audio influence maybe too low at 25%

[quiet solar]

Yeah, I've noticed Suno can be a bit funny about starting duets with a female singer. Give it a couple of attempts, and it should play ball on at least one of them. Then you can use that as an audio guide for further iteration.

Melody changes will be a combination of 'too low' audio influence and/or 'too high' weirdness. If your audio influence is 25%, I'd try nudging that up to 50%-60%. You don't want to go too high if it's reference track is a male singer that you're trying to convert to a female one.

Broadway musicals commonly use harmonies and ensembles in the chorus (it's what I did the tests on, as it happens), so that's a real bonus for what you're trying to do. However, you might find harmonies creep into verses where you didn't ask for them, or if you have lots of singer changeovers then some of them won't work correctly. It's just how Suno is - we can steer it, but we can't truly control it.

Depending on how much time/credits you're willing to invest, you may want to consider compromising on minor adjustments to melody and arrangement if it gets the singers perfectly right. Or vice versa.

[sweet ivy]

@quiet solar It really is amazing what Suno does. I used the 4 minor twice in the song, and it's a poignant song. Suno added it in the intro intro and outro, to compliment the lyric. If they had sung the last line correctly and not mangle the word "ever". I could have used either version. I could try another pass with audio influence at 50 - 60, but I'm worried even if they sing the correct melody on the last line, they may still mess up "ever", sounded more like "every"?

[quiet solar]

I'd say only worth being concerned about if you already know that word is a regular problem, or if you try a few times and Suno trips over it more often than not. Sometimes there'll just be a random word that gets mangled for no obvious reason.

If it is a persistent problem, you can try framing it as a phonetic alternative. Ev'err would probably do the trick.

Alternatively, if you're already comfortable working with a DAW, you always have the option to split out a perfect instrumental stem if you get one. Then you only need to worry about finding a fix for the vocals from other generations you make, which you can use to replace that part of the vocal stem.

That's been my go-to approach for quite a while.

[sweet ivy]

I do have to go back to the style prompt and tone down the ensemble note. On the final chorus it sounded like the end of Les Mis, big power note, key of C and a few women hitting the E above high C. Sounded incredible, but wrong for the song.

Getting late, will rework a few things in the morning, it is weird though, where the meter in the verses is busier, the singing was right on. On line they sang wrong, it's a simple melody?

[quiet solar]

I wouldn't read too much into it. Suno isn't intended to be a musical cloning tool - some degree of variation often creeps in.

If you need a really strict level of control over the final result, that may require editing rather than just tweaking prompts and hoping all the random factors fall into place perfectly.

[sweet ivy]

Yes, I understand that, but in the last 2 songs I did, Suno followed my melody perfectly. I think I can get the right melody, with an increase in audio influence. Once I get that, I can build around it. The entire song was on the mark except for one line at the end of the chorus.

[high snow]

I mean in the passed I have flagged and reported it in #🪲┃bugs-web but it’s like they skipped over it?? Idk. Bc it’s still a problem just like Heal Edits.

[high snow]

Tell them in #🪲┃bugs-web and also make sure to tag it on Suno.

[plain star]

I've already received your update. I still have 700 credits, but you won't let me make music. Pay first and then create! I'm fed up with this... Explain how this works?!!!(

[minor nest]

Please anybody have suno promo code

[quick wedge]

Suno doesn’t have any promo codes and I don’t think there’s going to be any unless there’s news of it in the server.

[quick wedge]

Your credits are still there, but you can’t use them to create music on Suno unless your subscription is active, so you need to renew your plan first; if it still doesn’t work after renewing, they should check their billing status, try logging out and back in, or contact support in case it’s an account issue.

Wrong channel to even ask that or you can contact them via email if you haven’t yet. This will tell you how to create one with the steps inside https://discord.com/channels/1069381916492562582/1373016771103166575

[solar lily]

I am new to Suno. I have a question about more control over instruments within a song. For instance, I'm working on a song where I want 2 cellos to carry the lead, with VERY subtle piano in the background. I described it similarly styles, but Suno keeps making the piano the lead. Is there a way to get what I am after?

[thick hornet]

Oh, I just "made" a song in their style. https://suno.com/s/saL0i205QBBAn4zN I spent two evenings, I really like the result.

[thick hornet]

try to put your wish in the track name as well, it often helps.

[quick wedge]

You can experiment with this prompt and see if it helps some

Cinematic cello duet with two cellos carrying the entire melody as the only prominent instruments. Cellos are fully dominant, expressive, and front and center. Piano is barely audible, very distant, and used only as a faint ambient texture with no melody or rhythmic presence. No prominent piano, no piano lead, no piano focus. The arrangement must remain cello driven from start to finish

[worthy kraken]

There is another isse that for some reason plagues all my songs.

When i try to extend on the second verse, it reads part of the lyrics, then stops, and starts over, duplicating the lyrics. But the lyrics entered is fine!

It seems like the model (v5) ignores all my prompt and generates like 20 seconds or something based on existing material, and only then applies the prompt. And i don't understand why. Maybe someone encountered it?

[polar spade]

We can often forget to just put things like "cello" , just "cello" with a comma after, , as our first prompt in the styles box. That can avoid any confusion for Suno of trying to explain it in a phrase. But im assuming we anticipate also making edits later to move any cello interludes and such around when they're not placed perfectly as we want.

[solar lily]

Thanks!

[solar lily]

Thank you

[solar lily]

Thanks

[polar spade]

Not to ignore the obvious of what its doing, but if it doubles up, then we splice it back to just one iteration of the verse to move forward with the project. Yes, then, as a side interest, work out what the issue is so we dont have to keep editing the work.

You can put a line from just before the verse to help it recognize where it is in the piece (in addition to the timemark chosen). Since it pulls at least 3 seconds left and right of the timemark to blend an extension in, its my thought that its actually voicing a bit of the extention lyrics when it shouldnt. Suno also need to reference the previous lyrics posted on the original first Gen output as well, it seems, so the question is always, should we , when doing an extension, visually remove that part of the lyrics from the original song.

Which has the better effect? Leaving the whole lyrics in the original, or removing anything which we are then placing into an extension. Ive ignored doing that recently as I havent had any recent encountered problems, but back in 3.5 I removed those segments to avoid this doubling effect. You might test, just as a benchmark.

[quick wedge]

You’re welcome 😊 have a wonderful day too ✌️

[shy heath]

Is there a definitive wiki that has all the different tags you can use for various instruments? I ask because I see people just pumping out really good songs, but I'm struggling with trying to get one that I like. For example, I'll have lyrics and generate the songs, and listen to them all and I'm not quite fond any of them, so I keep generating. I've listed out my genres, and used the AI to generate the prompt. But are there things that I'm just not doing to get a song that I really like? Or am I just being too selective and waiting for the "perfect" song? Some people are putting out songs almost daily and they're good, so I don't know if I'm just not doing something that would really help generate a song that I really like.

I've been using Suno for a long time now and sometimes I've had to generate dozens before there was a song that I really liked. Sometimes it would only take a few tries to generate something I liked.

[simple briar]

Yeah I get what you mean. That actually makes sense with the 3 sec blend, it could be pulling in parts it shouldn’t and causing the doubling.
I think testing both ways (keeping full lyrics vs removing the extended part) is a good idea to see which one stays cleaner.
I’ll try that as a benchmark too.

[worthy kraken]

How exactly do you splice it back?

Putting previous line does not do anything to solve the bug. It just starts with that line, but only AFTER it will sing the fake lyrics that are not there

Looks like it hardcodes itself to put that part there no matter what i type in the lyrics at all

[solar lily]

You as well. 🙂

[simple briar]

yeah that bug is so stubborn 😭 feels like it just locks it in no matter what. have you tried rewriting that section completely or splitting it up

[worthy kraken]

Rewriting how? Change all lyrics? I don't want to change it since it is what i want. And even if i would, it would still play up the bugged lyrics first, and then play whatever i enter. The point is that it ignores your prompt completley and puts its own stuff instead first, and THEN it plays my lyrics.

And about splitting, what does that mean?

[simple briar]

ahhh got you 😅 yeah that bug is wild, it just ignores your lyrics sometimes. by splitting I mean breaking the verse into smaller chunks and feeding them one by one sometimes that tricks it into following your lines. have you tried doing it line by line?

[worthy kraken]

Like, i extend, and just type one line instead of whole song?
Result will be the same, first fake lyrics, then my one line, then suno will hallucinate the rest to fill the empty space.

[solar lily]

Just incorporated your prompt, along with a few notes of my own, and it created something quite beautiful. Thank you again.

[simple briar]

yeah exactly 😅 even line by line it still tries to fill in its own stuff first. it’s like it refuses to let your lyrics lead. have you tried using super short bursts and seeing if it at least sticks for a few lines

[quick wedge]

You’re welcome I’m glad it’s working out well for you if there’s ever anything else never hesitate to ask and they’ll help you out in here.

[worthy kraken]

Does not work.

[polar spade]

If anything is repeated, without any overlapping to where a part is missing, then place in edit UI and cut and splice where needed. This would no different than an vocalist running a couple of takes to get a line down correctly. Then we just splce the best one in.

The more Suno glitches and repeats lines, sequentially and not overlapping, the more choices we have if it sings any of them differently, at no extra credits.

A glitch, yes. Needing fixed. Yes.

So I gave a workaround to move the project forward , and a possible explanation of why it's glitching. Otherwise, each and very time, also click that report button and give feedback of the error.

[worthy kraken]

You mean in song editor? Sadly it is entirely ueseles because it makes huge volume spikes at the connection point between changed segment and the rest of the song. Those are very audible... And for some reason remain even if you extend the song past it.

[polar spade]

Ah. Sorry to hear that. That does limit the options. Other than seeing if someone has a guarantees work around, then it remains frustrating. I get it.

Further testing:

• I mentioned removing the very lyrics were are using to extend, if you have not tried that yet.
• I mentioned adding the leading line from the original into the top of the extension to see if that helps.
• Another is to add [interlude] as uninterrupted placeholder, to keep it from needing to match any expected metering coming from the next verse. Again, if then ytou dont want the interlude, this would need removing it. Ytou say just doing that affects the volume. So it may not be the best to test.

OK. Otherwise, if we cant go further on this song and weve reported it as a bug until its fixed , what else would you like to see as far as a resolve?

[worthy kraken]

You mean removing lyrics at all? Leaving it blank?

The second suggestion i already replied to, it sings the wrong segment and then starts normal one starting from that leading line

Same with anything else you input. So in this case, interlude will happen after the wrong segment.

[polar spade]

As for the first, removing lyrics.... we remove just the lyrics that you are replacing or rather, those that are starting the new extension,, we would remove them from our original output. Probably half of users do that anyway, just highlight, cut and paste into the extension, but it helps to bring it up.

Again, its still unknown what the issue is. But just from my experience I get to make some guesses .... and to suggest what ive seen as promising.

If you come across anything that clicks, by all means, let me know.

[polar spade]

Ill add, if you remove ALL the lyrics from the original, the extension will go wonky and suno will often just sing a new song. So, no, you're correct to question that. We do NOT remove ALL the lyrics from the original when we do an extension from it.

[quiet solar]

There isn't a fixed list of tags - you can put in whatever you like and Suno will attempt to look it up (with varying degrees of success). Some tags may be more reliable than others, but that's sometimes you get a feel for over time and may change from one model to another.

In terms of feeling the songs your prompts are generating aren't landing where you like, what are the main differences between what Suno is creating for you and for others that you do like? If they're kind enough to leave their prompts visible, have you looked at what might have been different about the prompts they used that you could learn from? This is such a personal thing as to what sounds "good" or not that I don't know if we can really give you any suggestions without more info, but there is certainly an art to learning to steer Suno in the direction you want.

Personally, I find it's pretty normal if I'm creating something from scratch in Suno for it to take 6-8 attempts to get something I definitely want to continue building on. Equally, sometimes the first attempt is really solid and the one I keep on working with.

[worthy kraken]

If you leave at least a bit, it will keep the same pattern.

I feel like issue is that the model for some reason inputs part of the original generation first, and then proceeds with user's input. As to why it happens, i have no idea.

Seems like putting an extension very far away from that segment (like a minute before) fixes things, but you need to generate most of the song all over, and but probably might appear again. Appeared twice for me already, yesterday and today, in different songs.

[polar spade]

Thats been a good idea. If you mean putting some line spaces, maybe things like [pause] or some other tags to keep the extension lyrics physically further from the top of the lyric box. Thats a go-to for me, at times. 👍

[worthy kraken]

Nope, none of the tags work. Whatever you input happens AFTER the bugged segment. So you can't delay it, i already tried various tags and lyrics.

It always starts with the bug, and then proceeds with your prompt

[polar spade]

So, to open up being able to splice things in Edit UI.

Moving onto that other issue you metioned: What exactly happens in regard to "volume spikes" on a splice. Can you post or even DM me a screenshot so I can see whats up. Is it a literal short clicking spike or a quick spike and continual volume change?

[arctic cradle]

Hey guys I’m a drummer, are there any musicians in SC

[worthy kraken]

Sometimes it's like this - https://streamable.com/sh1e82
0-06, like song was cut and connected with other piece, melody breaks
Sometimes it's volume spikes https://streamable.com/h9j8kd (also 0-06)
You can clearly hear how song become much more quieter

It always happens in segment where the new (replace) generation stops. Seems like it is not being able to make a seamless transition with the rest of the song.

I actually have plenty of those examples, wasted so much credits on that... if i knew, i would not use the editor at all. Seems like it make a good transition only if some kind of a miracle happens, very rarely.

[solar lily]

This is something I would not have thought of. I'll test it out and see what happens. I'm also testing pushing my styles influence up to 100 (from 80) to see how that changes things. Thanks again.

[worthy kraken]

And it works both ways btw, on first example the transition is initial (from the song to the edit), and on the second example it's from the edit to the rest of the song

It is dull on both edges >_<

[polar spade]

By your description it definitely seems to be a thing. Something specific, in fact. But i can not tell just from the audio im hearing. So my ears may not be all that great.

Since you have this still in Suno, would you just pop it into Edit UI and taked a screenshot of the waveform, zoomed in a bit, where the edit splice you made, to help show and focus on the spike or volume change? And this volume issue is both directly withing the playback with in Edit UI and in Library?

The further youve described it, it seems your also saying that even the extensions themselves have a difference in volume, with no further editing or cut/splice involved?

[worthy kraken]

It is not super noticable in the volume, but here it is.

[shy heath]

Thanks for the response! I have tried putting in various tags with varying degrees of success, as you've said. What's frustrating is that putting in a tag like "[Rock Guitar Solo]" may or may not have a rock guitar solo. I can generate several songs with that tag and not all will have the solo.

I guess I'll keep experimenting with prompts and just see what sticks or works. Unfortunately some of the people I follow remove all all their tags and just have lyrics displayed. Maybe I'll reach out to them and just see if they have any tips.

Yeah, I get that, but I guess I just keep waiting for a song to really blow me away, which has happened. But for example, I have a song that I've been trying to generate something that I really like, and I have a 143 attempts currently. And those are just the ones I kept! I've probably deleted a dozen or so more, ones that right off the bat were terrible.

But thank you again for your insight. I guess there's no really right or wrong way, it's just whatever the AI generates lol. I know it's similar to AI generated pictures or videos too, that an AI can generate wildly different pictures from the same prompt.

[worthy kraken]

It is supposed to go down slowly, but instead it makes sudden spike right after the coursor

And then proceeds

[polar spade]

That curser line or playhead i see is indicating an edit or splice point? And my yellow line is your spike?

[worthy kraken]

The coursor is just before where it happens.
Yes, yellow is the spike.
But it always happens at the connection points of the edits

[quiet solar]

All tags in the lyrics box are inherently a bit hit or miss. If you really want a guitar solo, you may better off referencing that in the style prompt as well as just tagging for it. The style prompt is the most powerful influencer of what Suno does.

[shy heath]

Oh I've definitely done that! 😄

[polar spade]

To recap follow the line of discussion, you're saying this kind of spike occurs very often just after a splice point of any song, not just the same song? Ive not seen that before. Certainly interesting.

Since its in the waveform itself, and I see a slight spike just before the splice, we can speculate that Suno doesnt get a good crossfade every single time and those spike after the white line [cursor line] is just Suno trying to make it work.

The only suggestion then is to do something thats a pain, but may be worth testing:

Between verses Lets says we're putting an extension of the second verse after the first. Scan the musical space between the two verses to find the lowest near-zero crossing point, where the waveform has a low spot to set your timemark. The attempt is to give Suno something low to come off from. Next best, or better, may be to find a relatively mid-range part of the waveform to assume Suno can average the volume of the extension off of that.

From cuts and splices Same idea. Often we have enough from a doubling or overlap that we can find a low spot to cut from. This is typical in editing anyway, if its not a solo drum beat (which them we might splice on top of a beat).

If this spike is also happening JUST AFTER, a microseconds after, a cut though , and its consistent and feels like a Suno thing, then thats totally strange.

But what may be happening is the cut is at a split where the cut removes a nice crossfade and we get a punchy part where the continuity has been removed and the loud part of the segment on the right butts up against a less-loud part on the left. This again, would take zooming in close and finding a good zero-point where we can make a good cut so the two parts have a smooth join.

I think , honestly, from your description and screenshot of the wave form, we're talking more of the first issue, Suno not making a good crossfade of blend when it tries to add the extension. So test, but meticulously finding a perfect low-volume or zero-cross point as the timemark to extend from.

Hope this is helpful. If im saying stuff you already know, just tell me. Im just covering all bases.

[shell loom]

https://www.youtube.com/watch?v=K5j47JWr9_g&t=11s

[polar spade]

Im not clicking it, so whats this? In suno-school.

[shell loom]

how to make a script for long music videos, using suno and grok.

[young geode]

No, there's not an online resource you'll find.

However I have an extensive list of tags, many of which I have heavily tested and know work near 100% of the time.

The trick to SUNO Metatags is to just use already recognized musical terminology, the more recognized and established, the better.

Additionally, genres play heavily into the effectiveness of your Metatags and their strength of control: Funk songs just won't do Blast Beats.

Depending on what you're looking for tags wise, I can send you my list or you can look through my Suno profile and browse my songs. I leave all my tags and edits in for others to review.

[worthy kraken]

Yes, splice appears in any song. I remember it made seamless translations before, in past year, but in this it seems broken.

Yes, you had the right idea about making the split in the part where it will be least noticable, or where the Suno will have more freedom to add its own thing, like when the song pauses for a second without music or lyrics. But such poits exist not in every song, and the point of edit it so edit certain area that's bad and keep what is good. If i would have to remake everything up to certain points of silence, it will defy entire idea.

And yes, you are right, it is the first issue. But not always. Sometimes it even inserts very different type of music into the cuts, like, totally alien sounding, like it cut one track and added a random one into it.

[polar spade]

Suno Tip Small gen inside the extend tool. Zooming into the timeline of the waveform.

HI all. Just a quick vid of a small tip some may not know of. Helps with extensions.

[young geode]

To give you an example:

If you wanted a guitar solo, [Rock Guitar Solo] will confuse SUNO, as Rock is a genre, using [Solo] would work better.

However, if you prompted as follows:

[Instrument:Guitar, Stratocaster, Djent, Power Chords]
[Uptempo]
[Release]
[Scale: Phrygian mode] 
[Solo: Guitar, Stratocaster, Arpeggios, hammer ons, Djent, muted palm]
[Solo: Guitar, Stratocaster, power chords, Djent, muted palm]
[Solo Ends]```


You'll get a much better result.

[polar spade]

Then at that point, we've gone over most ideas others and myself have used before. And left with it being a reportable bug/glitch.

No harm bringing it up at another time and see if anyone else has anything new to add.

[shy heath]

Yeah, if you can share your list of tags, that would be great. That makes sense about my "Rock Guitar Solo" tag, as it works about 50% of the time. I've been on a several-year 80s/synthwave kick with sax and rock guitar solos (though the sax sounds pretty terrible in v5, I've been mainly sticking with guitar for solos).

[young geode]

https://discord.com/channels/1069381916492562582/1382019029266600067

The files are contained in the last message of that thread.

As for what you're trying for, these are songs I've made that have a similar profile:

https://suno.com/s/L1nMk344f1T6kB1j

https://suno.com/s/oDaHes6lCXq1m80q

[shy heath]

Thank you! Much appreciated!

[shy heath]

You mean in the comments? It's showing no comments for me.

[young geode]

Yes, but I'll just send it directly.

[runic rose]

didn't know where else to put this but here we go

If I subcribe for this month and don't use the credits, shdnt I have them available when I subscribe the next month?

shdnt they be added to my next month subscription credits?

[frozen dagger]

they do not rollover.

[runic rose]

would be nice if they did though

[frozen dagger]

if you purchase top off credits. you keep those, but sub credits don't rollover. (nad need an active sub to use the credits)

[umbral swan]

Hi everyone! I’m looking for some help with my Persona. I created a Persona based on a song with a voice that is absolutely perfect for my project. However, the original track has many vocal 'runs' or high-note belts that the AI now replicates in every new song.
It's becoming a bit of a struggle because I want to keep the core voice but get rid of those specific vocal habits in new generations. I managed to generate one song that sounds much cleaner, but I noticed I cannot create a new Persona from an existing Persona-based track.
Is there a way to 'transfer' or migrate a voice to a new, cleaner Persona? Or is there a trick to suppress these vocal runs in the 'Style of Music' prompt while using a Persona? Any advice would be appreciated!

[frozen dagger]

One way would be to open the track you used to create the Persona, cut out the undesirable parts, and create a new Persona without those behaviors.

Alternatively, if you want to use one of the songs you made with the Persona to create a new Persona, first open that song in to the editor, save as new song, wait for it to fully load and the option to create a new Persona will be available.

[umbral swan]

That sounds like a good plan! I’m definitely going to try this out.
Thanks for the tip, I really appreciate it!

[umbral swan]

@frozen dagger That is absolutely genius! It worked perfectly. I tried your suggestion and it's exactly the fix I needed to get the Persona back on track.
You’re a legend for this tip, thank you so much!

[frozen dagger]

😂 You're too kind. I am glad it worked for you.

[umbral swan]

I’ve been searching for a solution for two weeks now and had no idea it was actually this simple.

[frozen dagger]

Wealth of knowledge being shared on the discord. Lots to learn still. Things are always changing as well.

[umbral swan]

I do my best to keep up with you.

[left hornet]

Im so glad that i randomly checking this channel and found this tips, its really improve my persona a lot. Thanks for sharing.. It fixing so much about screaming things, now my persona can sing with calm and joy

[stable sandal]

)What is a Persona, how is it used, and what is it good for?

[tight prism]

A persona is a locked vocal persona so you can have consistent vocals in all your songs.

[dusky lark]

Does anyone know how to control the model so that the two chorus, which have exactly the same lyrics, turn out to have exactly the same melody?

[tight prism]

What are you doing now? Generally the chorus does have the same melody...

[hollow adder]

🎬

[polar spade]

Nothing stops us from first trying to duplicate the chorus and pasting it in the next spot. Have you tried wiggling one in to see if it blends in and sounds good?

[worthy kraken]

Well. I am exausted. My lyrics gets the duplication bug in every single song that i try to extend. The only thing that fixes it is starting far away from that point, but it is not possible very often.

Maybe i am doing something wrong? Can't such a crucial bug appear that often?

Even on 5.5 it persists.

Like, look at this. https://streamable.com/a06t8b
Read the lyrics in the extend section. Is it anything like what is being pronounced?
Yeah. It literally takes the ENTIRE VERSE from the beginning, and puts it there in the extend, and plays it before doing my prompt.

... why?

[stuck junco]

Hey Void! Could you share a song link for this here or to me via DM?

[worthy kraken]

I need more clarifications regarding which parts do you need. Wrote it DM.

[stuck junco]

Thanks, reaching back now

[polar spade]

Is "This is hoper 5..." the first line in the original? So I can relate to where we are in the video.

[worthy kraken]

Yes. It is the first line.

But if i will extend from the different time, it will input different lines in there. For example, i have another extend, that inputs there lines from the future, that was not spoken yet.

Like it generates random part of the song (often with hallucinating lyrics that are being cut or repeated) and pasts it ahead of what i asked for.

And this time changing timing does not heal it, just leads to different hallucinations.

[worn thistle]

hello, I'm sorry to ask you guys..... but i have a problem with upload.... i can't upload....... it allways said "upload fail"..... like 20min in row.....

[stuck junco]

Hey depito12! Thanks for flagging this!

The team is currently looking into this. Thank you for your patience and understanding.

[worn thistle]

ok, ty very much, and, i have a question, i"ve been ind, two months out of this server and it have change alot..... sooo, before there was a room for discussing tags and personas and bugs, and the comunity of this server tried help others..... where i can find asimilar room? is it here or somewhere else?

[stuck junco]

Happy to help! And welcome back!

A great place to start is the #✨┃getting-started channel 🙂 That will give you a quick guided tour of the server layout, but if you have any questions, please don't hesitate to reach my way whenever!

[worn thistle]

Thank you very much 🙂

[polar spade]

Since its pulling lyrics off of past generations ("future" intended lyrics) I'd like to ask you to do these steps as a test in this order.

1. Save (bookmark - and check to see bookmark has been saved) or copy whatever prompts you wish to use.
2. Click the trash bin to delete and clear Create panel.
3. Click your original (to extend from )
4. highlight and cut (not copy) the "second half" of lyrics to be used in extend -- only leaving the first half of lyrics in original
5. Click edit/extend of Original
6. type in :53 seconds in timemark of the extend box in Create panel, and not use slider
7. paste in second half lyrics
8. add prompts , sliders ,etc and click Create.

Tell me what happens from that.

If you've done exactly these steps before, then ignore the request.

Any idea why the squiggle lines under all the lyrics as if auto-correct spelling is needing a response? Thats interesting.

[worthy kraken]

I am stuck at the step 4, cut from where, exactly? The lyrics displayed on the right side of the screen is not editable.

[worthy kraken]

Dunno about that, maybe firefox just do not have the spell checker or something

[polar spade]

Let me know if that made sense.

[worthy kraken]

Yes, but it's called edit displayed lyrics, i think that this feature is only for when you publish and does not affect the song. But i will try that now.

[polar spade]

Youre right, that it shouldnt do what its doing. Only from experience with 3.5, Its doing some familiar things as before, and this is a cleaner way to help it from pulling from previous generations.

[polar spade]

Editing the lyrics of a previous generation in the chain of iterations will ( at least from past models and history) actually affect the future generations, such as covers and extensions. If present model have improved that glitch, then I'd hope I'm wrong now.

[worthy kraken]

Nope, exactly same but, it starts song from the first verse (but with different generation), then hallicinates for a few seconds, and finally proceeds with my lyrics

[worthy kraken]

The problem that i only have one iteration (original generation) that i am trying to extend for.

Yes, this is that bad, now it does that right away.

Btw it is curious if it's bugged for everyone or just my account. So if you will want, i can dm it to you and you can try extending it yourself and see if that will work.
Maybe that will fix it somehow... or maybe not.

[polar spade]

OK. Thanks for the respect to try the test anyway. Its not working the way we should expect it to work. The hallucinations can now be coming from the fact its looking for the lyrics its suppose to "replace" but you've removed them.

So . . . and this is just me .. . but I have gone back to the original , first, generation before making any future covers and fiddled with lyrics there. But it shouldnt be this bad to have to do that.

The other suggestion is to make a copy of this "original" song , the one you're working on now -- the one with the orange flame. So that is literally has NO generations before it to pull lyrics from. Then extend from it however you wish and think are the cleanest steps.

Again, I've done lots of extensions in the past year, and this has not been a thing recently, leading me to think it should not happen anymore. But that has seemed to be the reason with up to model 4.5 -- is that its pulling from the archives of lyrics lurking in the chain of iterations before this one.

[worthy kraken]

No, it's on 5.5

And was on 5 yesterday

I can copy a song?

How?

Remaster? This is the only thing that comes in mind.

[frozen dagger]

@worthy kraken i saw in the earlier video you were working from "My Workspace", going along with what @polar spade is saying about the previouis gens. I would also create a new workspace, move your original track to a new workspace. Again, this is not something we should need to do, but when things start feeling wonky I do start a new workspace and try to "start fresh" in addition to the things Dasvon is already saying.

[polar spade]

Always pointing to the usual. When the thing starts glitching from what is expected , then start over and at minimum, clear localStorage, close browser, open browser. Never hurts to keep the slate clean. As you've seen in the announcements they're throwing in new updates as we speak. Any old client files or old files in storage need to match up with what the Server is looking for to make the visual and some audio to sync up better.

[frozen dagger]

Create a copy by opening the song in editor and save as new song without making any changes.

[polar spade]

My comment Removed. We'll go with the easier route. 👇

[polar spade]

@worthy kraken yep... while in New Edit UI, just click Save As New File. Easy peazy... do it that way.🤘

[polar spade]

yes . Correct. After 5.0 came out, this should not be a problem. But you're showing the issue has raised its 9 heads and striking again. So, we hope something is glitching and resolves itself. For now, the suggestions are more about keeping the route and steps as clean as possible and away from involving too many previous iterations in the chain.

[north scaffold]

Are there still no loops in suno?

[ripe galleon]

can v5.5 do a jaw harp

?

[hollow jay]

So v5.5 is a downgrade, you can’t make personas from existing tracks that are not your voice anymore….

[ripe galleon]

what if you put on a funny voice and try making a voice model out of that and see if it accepts it or not

-# idk how personas exactly work

[hollow jay]

Not that

I mean suno generated songs

[worthy kraken]

So i made a fresh copy, put it into new workspace, but got exactly same issue(

[hollow jay]

So if I generate a rock song and I want the voice, I can’t make it a persona

[ripe galleon]

hmm what if you like,

download the song

isolate the voice

and then upload it as a voice

?

do you think that could work

[frozen dagger]

haven't tried it yet, but it says Voices is Persona now, same funtionality, new name. All your Persona should be in Voices tab now.

[worthy kraken]

I guess the only one think i haven't tried yet is to let someone else extend. So if anyone is up for that, write me.

[hollow jay]

When you select song from library asks me to verify the voice

[ripe galleon]

verify?

i dont think ive heard anyone mention verifying the voices before

damn

[peak lava]

In 5.5 Suno doesn't use the BPM that i ordered, and in INSPO, SUNO copies parts, not inspires. I'm the only one?

[polar spade]

it's borked. 😜

I could use my own song, but would you mind to DM me your lyrics as they are arranged and a couple of prompts. And I'll literally create an original and then extend from around 53 seconds for the second half. Lets just see whats up with when using a different computer.

[hollow jay]

That means personas been Suno is dead man, they killed off creating personas from generated songs, bye bye suno rem

[peak lava]

damm. for me it's so bad news

[hollow jay]

It’s game over for me

[worthy kraken]

That will probably work. The bug does not happen with every new song. It just appears eventually, after many edits.
But this is the first time when it appeared without the edits.

That's why i wanted to share the song that you (or someone else) could try editting for it to see if it will be bugged for them as well.

[peak lava]

maybe for me too. i used personas for a lot of songs.

[frozen dagger]

I just made a Voice using an old Suno track from my library, process was same as Persona, didn't ask for verification.
If you're re-uploading the Suno track, then no can't create a voice. But shouldn't be an issue since it's uploading a track that is already in your Suno library?

[polar spade]

You can make personas from suno voiced generated songs.
Nothing has changed as far as what we could do before. Voice is the new name.

[hollow jay]

Using the new voice option?

[hollow jay]

There is no option …

You guys even using the new interface

[frozen dagger]

from the Suno track, yes. click the 3 dots (right-click) create voice

[hollow jay]

Don’t see option, I went to voice + and ticked all requirements and library and asked me to verify voice

[royal moth]

Where is the option to train/upload your own voice?

[frozen dagger]

We have never been suipposed to use uploaded audio to create Personas. We have always been able to use Suno generated track to create Personas.
The onlt change is Persona is called Voices, and we CAN use our OWN voice from upload (with verification)

[half dragon]

Custom Model in the dropdown

[royal moth]

Thank you

[frozen dagger]

For creating Voice with a Suno generated track.

[quiet solar]

Can confirm creating a voice from a Suno-generated singer is exactly the same as it was before, using Create -> Voice from the Library. Works just as it did before in new songs.

[polar spade]

The weird feeling of it though is I can now make a voice from an uploaded song of my singing a song, but I can't cover my singing a song with Suno and then make a persona/voice of the female covering it as its an uploaded audio. lol. Same as before, but now just rubbing my nose in it. 😆

[hollow jay]

I went to voice clicked + ticked the checkboxes, select from library, selected track, then asks me to verify voice

[frozen dagger]

this is only for creating a Voice using your own voice, not a Suno generated voice.

[hollow jay]

I understand so where’s the suno generated voice

How do I revert back to old way of creating personas

[frozen dagger]

It's same as before select the song from your library open the menu for options it's under Create > Voice
Same as before except it was called Persona.

[tight prism]

You don't.

[hollow jay]

I don’t get the option there

[hollow jay]

Ok, but the voice option isn’t available on create either just on the create popup

[frozen dagger]

If it was an uploaded audio you won't have the option. We have never been able to use uploaded audio to create Persona.

[hollow jay]

It’s not uploaded, it’s generated suno

I’ve been making personas for 6 months

It’s all via generated suno tracks

[frozen dagger]

u dont have these options when right-clicking your song (or clicking the ...)?

[hollow jay]

Yeah I don’t for some reason let me check again

[tight prism]

You are Pro/Premier?

[hollow jay]

Yes

[frozen dagger]

I haven't had a chance to test for some of the other things that can prevent it, but that would only apply if you have the option there but it is greyed out.

[hollow jay]

Let me generate another song and check again, but my already generated tracks don’t show it

[half dragon]

Maybe UI updates are not reaching everyone yet?

[hollow jay]

You’d think it would be universal lol

[quiet solar]

If they're not from uploads, that sounds like a bug. The option is there for me on any generated audio that wasn't made from or derived from something with uploaded audio.

[hollow jay]

I’m aware of not being able to create personas from uploaded audio, but as stated these are all generated tracks

[frozen dagger]

good catch on the derived from, i shouldve included that as well for clarity

[hollow jay]

I just generated a new track, normal, no voice option

[quiet solar]

Yeah, it sounds like a bug. Try using a different browser or resetting the cache.

[frozen dagger]

can u screenshot and show us the sub menu option it shows you?

[hollow jay]

Wait, I just noticed, I was using generating a track with a persona, and it won’t show the voice option on create, but when I remove my persona (then generate) I see the create > voice option

I used to be able to make a new persona from persona generated track

[polar spade]

Thats been awhile that we now cant stack a persona from a persona'd track. The menu choice of "Make persona' or "Make voice" does not even show on those cases now.

[hollow jay]

Previously I didn’t c the option but when going to create popup and personas it did allow generating

[vague stratus]

I have an cinematic ambient track I'm trying to extend, if I just use extend it gives 5-17 second extensions. only way I got it longer was to type a lot of lyrics, which I don't want. any other way to guarantee longer extensions?

[hollow jay]

Thanks yall

[hollow jay]

Thank you

[hollow jay]

Thank you

[hollow jay]

Thank you

Issue resolved

[polar spade]

Thats cool if it appears later or should be that way. I;d much prefer to make personas from seasoned personas songs as before.

[hollow jay]

DM, I’ll guide you

[pure panther]

Some cool stuff just dropped! I hope I have time to play with it soon 😄

[queen linden]

Ok, educate me on a good microphone...lol, I have absolutely no idea what i am looking at 5.5 looks like it can be a lot of fun

[hollow jay]

IPhone

[thorny shuttle]

I finally started generating the EDM sound i love however the format is always short intro and outro's, Ive been working inside Stuido, Steming however everytime i stem i lose chunks of quality, is there a fix for this or is that the way it is?

[timber forge]

So how do you make suno create a song with 2 male voices? It keeps failing to do it. I have no problem with male-female duets, but this doesnt work

[worthy kraken]

Update for my issue - i even tried to make a remaster of that song, still bugged( Looks like that generation is simply cursed

Sems like i will just have to yet again forget all the spent credits and try anew...

[hollow jay]

They’ve still not fixed the persona sounding different on instrumental cover than without instrumental cover

The voices are totally different

[quiet solar]

You need to have two very distinctive voices; try something like prompting for one to be a tenor and the other to be a bass-baritone.

[naive seal]

ok, what happened to the personas now?

they're called voices? -.-' not appreciating that change

[frozen dagger]

#📢┃announcements message
Persona is now called Voices, same as before with adding your own voice as a possibility now.

[glacial cipher]

its not the same tho, you can make a "voice" from a song anymore

[frozen dagger]

The Persona functionality is still there. Just called Voices. Select the Suno song you want to create a voice from and right-click. Create -> Voice

[frozen dagger]

Is that a song that was made using a Persona?

[glacial cipher]

thats a song im trying to make a persona from

[frozen dagger]

still to verify, did u make it using a persona? or an upload?

[glacial cipher]

persona

nvm i see it now

you good, human error

[frozen dagger]

If you made the song using a Persona then it has the Persona attached on it. Open the song in editor, save as new song, this creates exact copy of the song without the Persona icon thing so you can now create a new Persona from it.

[vernal hound]

Yo is this like a sort of support channel? Cause I cant find support anywhere?

[pure panther]

#ticket-help

[vernal hound]

Thanks brotha

[glacial cipher]

ohhh ty

[pure panther]

You're welcome 🙂

[vernal hound]

Maybe its something the knowledgeable folks here can hel with?

[pure panther]

Perhaps, feel free to ask

[frozen dagger]

no harm in asking, but we are users helping users. if it's like tech support related to billing or your account that is ticket help

[vernal hound]

I made a song on the free version, it gave me 2, V.4.5's and 2 V.5.5 previews, It said "Upgrade for full song" So I did and paid the subscription, now it still says the same "Upgrade for full song" even though I have the 2500 credits from the sub?

I guess its like an account or billing query, I'll open a ticket also why not

[frozen dagger]

I would refresh the webpage first, but not expect much, but possible. Sounds like could be a delay in payment processing? How long has passed?

[vernal hound]

probably 5 minutes, the credits came in instantly so thought all g

but

Ok was deffinitely a server or pay processing delay, we all gucci now

Got the full track now

Appreciate yous willing to help 💪

[obsidian frost]

is there any easy way to get songs to last longer?

it seems like in the new 5.5 my songs are all under 2 minutes

[vernal hound]

Try "extended" tag or something like that, I honestly have had some songs come out at like 5+ mins but used "slowed" tag

My experience was older versions tho

[frozen dagger]

add some tags to include more music inside the lyrics prompt box by adding things like [instrumental drop], [interlude], [breadkdown] etc. Something between lyrics (or at the start and end) to put more music in

[obsidian frost]

will try, thanks

[lone flame]

Add either 5 minute duration or extended long duration in the style prompt. I regularly generate 6 to 8 minute tracks.

[obsidian frost]

thanks 😄

[lone flame]

Sometimes tracks are so long Suno doesn't know how to end them so they just cutoff.

[pure panther]

You can usually get Suno to add an instrumental section or solo by doing something like this:

[Break | instrumental: guitar solo]
..............
..............
..............
..............

That can help extend the overall length

I usually add two or three of these in my songs

guitar solo is an example

[lone flame]

Anyone know how to get Suno to use a word or phrases as samples throughout a track at random?

[pure panther]

You could ask it to randomy add a phrase in the style, but I think that would be very hit or miss.

[pure panther]

This is what Suno chat said about it

[frozen dagger]

if it is just an instrumental track and I use a Voice with what I want repeated, and no lyrics in the lyrics box.. maybe. It would probably pull the word at (quasi) random places.

[lone flame]

Thought maybe a command like use "this" as sample and syncopate it with the rhythm, or something like that.

[fast plume]

Where do I go for help?
I can't remaster a song because it has an artist tag and tells me to change the tags but I don't know how to change them.

[frozen dagger]

closest to that would probably be what @pure panther is suggesting. I'd use the chat maybe too if you have it, but don't know if that would work or not.
I usually make weird sample voices in the "Sounds" and duplicate them in studio to play with using them in that way, but the randomization part would be kinda tricky.

[frozen dagger]

where is the artist tag? in the style description?

[fast plume]

I think? I dunno about exact naming but it appears below the name.

[frozen dagger]

yes that is in the stlye description. So, open your Song Details. And the 3rd box is Add a stlye summary. Add your style summary there. (you can copy poaste what you have and just remove Daft Punk inspired), Save.
Open song in Studio (I think editor works too) and save the song as new song. This new song will be an exact copy with the new style description.

[fast plume]

Ohh, I'll try it. Thanks 🙂‍↕️

[frozen dagger]

GL

[frozen dagger]

Just tested. Editor works just fine.

[split jewel]

Good morning. I would need help: with the new version released today, I'm trying to clone my voice (with a very good recording of my voice for 4 minutes), and the result doesn't look much like my voice. is it normal?

[pure panther]

Did you increase the audio slider in advanced options?

it's set to 25% by default

[split jewel]

Hello, thank you for your reply. Which slider?

[pure panther]

🙂.

The audio slider on the bottom

[split jewel]

Ah the audio influence? Do I owe the best to 0?

[pure panther]

The higher it is, the more it will match your voice

[split jewel]

Because currently I put it about 90% and the result is not good unfortunately

[pure panther]

Ahh

[quiet solar]

It's still new for all of us. But a couple of basic tips would be:

• Make sure you've got a fairly decent vocal range in the recording
• Increase the audio influence slider
• Sing something from the style of music you plan to use in terms of genre

[frozen dagger]

I take that back. It operates differently. If done with editor it saves it as the summary only, the original descript is still there, if saved as a new song in studio however, then only our own style summary is there.
I hope still get what you need, but if trying to do this wiith editor still doesn't allow the remaster cus of the "Daft Punk Inspired", then Studio save is the only option with this method I am aware of.

[split jewel]

Okay thank you. And by doing this, my voice will be authentic?

[pure panther]

That's the aim. Your mileage may vary.

[fast plume]

I'll have to retry some other time on PC. It's not working on my phone. 😞

[split jewel]

Okay thank you. Does Suno also take care of the vocal effects directly during the generation on my voice? Because I find it rather raw and without effect or autotune

[quiet solar]

I presume Suno will create an AI voice based on your own, and at that point it'll operate like the AI voices in standard generations - so when it applies production effects on the vocals, it'll apply those effects onto your AI cloned voice.

[split jewel]

Okay thank you @quiet solar 🙂

[frozen dagger]

I finally got my voice in there... does sound like me... for better or worse. XD

[thorny shuttle]

can you do the same for longer intro and outro, im trying to get it to genrate mixable tracks for DJ, this is super interesting what you posted

currently ive been stuido editing and copy pasting to get both intro and outro

[pure panther]

Absolutely, just label it respectively, like for the intro:

[Intro | instrumental: guitar solo]
..............
..............
..............
..............

[thorny shuttle]

whats the dots for?

[pure panther]

Oof, yeah, I never do that haha

To get it to do it longer, more dots means it's more likley to go longer

[thorny shuttle]

wow ok, this opens a new world

[pure panther]

I put it in pretty much every song, it's very useful

[thorny shuttle]

thats amzing ill have a big crack at this, very useful

[frozen dagger]

nice. i thought your dots were just placeholder for the lyrics at first. nice trick.

[thorny shuttle]

i was stemming and using copy past to get the format correctly, everytime i do this i lose overall qaulity

[pure panther]

I first tried it after seeing, I think Ginger?(maybe Kanthric), mention using dot designs for long instrumental songs.

[tall vine]

so I noticed that it says that you can extract stems without FX, how does one do this?

[frozen dagger]

As far as I know that is only in Studio?

[tall vine]

I have studio I just cant find the option

[frozen dagger]

richt click on the track to open the sub menu look for Remove FX there

[tall vine]

thank you!

[high snow]

Hey, so I saw the newest update that they have for version 5.5, it looks great and all, but I was looking at the YouTube video and how they were showing the example for the custom model and looking at someone’s instrumentals audios. How do you make just instrumentals without lyrics? I know that’s a really stupid question but I always wanted to do that and just make instrumentals but I just don’t know how. And I don’t know what to put in the styles box and the lyrics box. And how to describe what I want. Can someone help me out?

[high snow]

Sorry, I honestly didn’t see that I must be blind as bat. Thank you.

[quiet solar]

Either leave the lyrics box blank, or use it exclusively to give directions and guidance on specific movements through tags, e.g.

[Intro - long sweeping flutes, soft synths]

[Verse - high energy sax solo]

Because it's instrumental only, you may want to make your style prompt a little more detailed if you want to steer Suno as to exactly the sounds you're after.

[rancid elbow]

happens to everyone eventually, all good

[high snow]

I wondered instead of doing verse three, etc. could I do like climax swell of cellos and violins increase/decrease for like if I wanted the middle of the song to increase and volume or decrease kind of like the best way to describe is like a climax basically and then you’d have the decrescendo.

[high snow]

I hope what I’m saying here makes sense.

Example of what I mean:

[intro — flutes, and Piccolo‘s and harps begin]

[climax — cellos, oboes and clarinets excel to FF]

Etc…

[quiet solar]

The fact you're using musical terminology is a really, really good start. The best advice I can give is to test things and see what does/doesn't work. Style can handle natural language (to some extent - don't ask it to do anything requiring context or logic though!) so literally describe what you're after and see how it goes.

It's really hard to say much in detail, because "How do I describe what I want" is basically the ultimate goal of prompting, so it's too broad a question to answer without knowing more specifics about what you're trying to do.

[high snow]

Honestly, the only reason why I know music terminology is because I used to be both in marching band and in concert band from middle school all the way to high school so I know little bits and pieces of it. Not like for conductor terminology, but it still comes back to me. 😅

[quiet solar]

Yeah, that'll help give it a steer for sure. Just bear in mind that anything you put in lyrics tags is typically less reliable and influential than anything you put in Style, so set out in style what you want across the song, then (optionally) give it a structure for how you want that to develop using lyrics tags.

[high snow]

Do you think I’ll listen to more prompts that I give in styles then in the lyrics prompt box? Especially if how I want this song to go and want to be added versus to take away?

[quiet solar]

Former quads and lead snare here, before they demoted me to waving a big stick and conducting the whole thing.

I'm not the world's expert on musical terminology, but I can make a solid amount of noise on demand. xD

[frozen dagger]

"commands" 😬

[high snow]

Sorry that was a typo. That wasn’t the way I wanted to use.

[quiet solar]

Yes. At least, for 4.5 and 5. 5.5 is too new for me to be able to say one way or the other from experience, but I imagine it will be the same.

I'd also echo OBLMOUS' grimace. It's best to think of prompts as guidance and suggestions to Suno, rather than 'commands' that Suno will definitely follow, or you risk walking into a lot of frustration. It's more likely to listen to what you put in Style than what you put in Lyrics tags, so bear that in mind for what is most important for you to tell it to include. 🙂

[frozen dagger]

I was waiting to see it echoed by Kanthric. It's just a frame of mind thing.

[high snow]

I don’t use version 4.5. Before the news version that came out today, I only really used version five.

[high snow]

Oh, OK, because the emoji that you gave me you made it seem like it was a bad thing 😅😮‍💨

[frozen dagger]

from what i'm seeing you got a good handle on things already. just was weery of thinking of them as commands, cus that they ain't. lol. Suno interprets at best, hallucinates at worst.
Crescendo should also be good for the increase in volume. There are other more detailed forms of it chatbot can tell ya, but in the same sense of Suno interprets, those may not hit with the nuance. BUT it is a new model. 🤔 No harm in trying things like "Subito crescendo".

[high snow]

Well, the only reason why I was thinking of crescendo’s, etc. is because of how well the new version that they came out especially what they showed us on YouTube. I don’t know if you saw the YouTube link and announcements. They did V5 (which was what we had now before the newest update came out.) and V5.5 (which is what we have now). And in the video they did a comparison between the two and the new one seemed to have done a lot better job with increasing volume or at least just how it sounds altogether. I mean, if it doesn’t do crescendo and all that honestly it’s fine. I’m just looking for more of instrumental quality now.

https://youtu.be/eVlFcpX1VGA?si=GifxcGRF4xw2Fr-B

Or at least a test out instrumentals. Because I’ve only done really anything lyrical mostly.

[frozen dagger]

I've seen it. I've been enjoying the new features myself so far. A few hiccups, to be expected.
Instrumental work is a lot of fun imo. I've been lazy and using the auto lyrics. 😂

[vernal hound]

auto lyrics actually went crazy hard for me earlier

[frozen dagger]

i'm diggin it. the auto-lyric box is it's own prompt box that i didn't give any attention to before. i see it now. for sure.

[vernal hound]

I sually only do instrumentals too

usually*

[high snow]

I wonder since they came out with this new version, is that why version five was so damn buggy and why there are so many bugs always arising and coming back week after week month after month? Because I kind of hope that was the case and I hope that there’s no new bugs that arise with this new version because I will cry.

[frozen dagger]

get a box of kleenex ready then. that is the nature of a.i. in general. this is all new. i just try to be grateful, and remember how I would have loved to have this when I was a teenager. (long, long ago)

[high snow]

I mean, I don’t blame you. It’s just that the bugs in general were just super frustrating. I mean it’s working for me right now so I am grateful.

[frozen dagger]

i expect we will have some kinks to work out, then it will (hopefully) be smooth sailing for a while. But I expect issues will also arise, as they do. Suno team been good, imo, for getting things straightened out. There are a lot of people using Suno.

[young geode]

It'll help you.

Metatags that use the musical terms have the highest and most reliable effects.

[high snow]

Tysm.

[high snow]

If you have any example prompts that you have like style prompts, is it OK if you could send them to me in direct messages so maybe I could get an idea of what and how you did it?

[vernal hound]

I get this isnt a support channel but what sort of bugs, like mine has created some bangers

[high snow]

Well, these were bugs in the past. It doesn’t matter now cause they’re smoothed out for now.

[vernal hound]

4.5 is solid im only trying 5.5 for the first time

and its doing a damn good job

[young geode]

Sent.

You can check my profile and everything in the last year uses that format. I should note that if you know the model of an instrument, and can fit it, it does aid in the sound.

[high snow]

Honestly, I have yet to try out the new version but I’ve heard so far that it’s doing pretty good and I’ve seen the YouTube video in announcements and it looks pretty fire

[vernal hound]

It seems to definitely follow the prompt better and sounds more manmade rather than AI

[high snow]

Honestly, exactly that’s what I’ve heard in the YouTube video and they did a comparison between the old version and the newest one that just came out. And it sounds a lot better. It sounds almost more unique in someway. I don’t know how to explain it, but sound wise. It sounds a lot better.

[vernal hound]

I mostly focus on instrumentals but, it csn actually make bangers

I've managed to make the most farty hard bass 😅 its wonderful

[high snow]

Did you just say farty? 😂

What’s that’s supposed to sound like? Actual farts?

Jk

[vernal hound]

I meant distorted 🤣

[high snow]

Ooooh 🤦🏼‍♀️😅🤣

[vernal hound]

Honestly the "distorted" prompt adds a perfect layer

[high snow]

How do you purposely distort your prompts? Like how does that sound?

[vernal hound]

extreme bass, heavy bass, industrial bass, distorted, slowed,

[thorny shuttle]

ive found exluding styles (im doing hard trance) like Psy, Big room and Uplifting help generate a darker harder sound - is anyone else using prompts like this?

[vernal hound]

I usualluy throw this at the end but the bass prompts work,

[vernal hound]

Have you tried hardsyle, sensation black, -- as tags?

[thorny shuttle]

im trying to avoid the clang style of hardstyle, generally hardstyle has a high hat on every hit and a shitting sounding kick drum, not all but most

[vernal hound]

oh sorrry youre mentioniong neg tags

[thorny shuttle]

more driving hard kick drums, im going to try your prompts listed above, but i use "pounding driving kick drum and base line"
"base line rolls into each kick drum"

[vernal hound]

man you are looking for a niche that AI can barely find

[thorny shuttle]

yes i am :p

working it to the core

[vernal hound]

you will have to be super specific with prompts

hahaha good luck

[thorny shuttle]

ill send you a priv link - added you up

[vernal hound]

You can emphasise the 808's maybe

[thorny shuttle]

yup i use 303, specific synths Virus so forth

[vernal hound]

but what kick you want to hit?

[thorny shuttle]

i havent ever actually specified a kick drum

[vernal hound]

Im just a blow in, im into bass hahaha

just trying to help man

[thorny shuttle]

its all good feedback mate, appreciate the info

[vernal hound]

Appreciate ya man! make some bangers ❤️

Someone provide this man answers

[hollow jay]

Can anyone explain why after 2 verses the persona voice goes weird, it loses the persona voice

I can’t make tracks with 3 verses 16 bars because of this, I can only do 2 verses

4-16-4–16-4 structure

[thorny shuttle]

ive got heaps to work on the new prompts and V5.5 so many new avenues

[vernal hound]

Have fun bro its new to us all

[frozen dagger]

@vernal hound @thorny shuttle you all tried experimenting with Sounds and Samples? I've had fun with bass music and using that.

[hollow jay]

It’s with v5 to

[thorny shuttle]

preloaded sounds? sample packs?

[frozen dagger]

Suno Sounds. and then use those sounds in the Sample mode (can be anything that does much better with short audio clips tho)

[frozen dagger]

Are you using your own Voice as a voice or is this a Voice that was made using Suno?

[thorny shuttle]

no never used that function to be honest,

[hollow jay]

Suno generated

Persona drift …

[frozen dagger]

Have you tried going to the persona page and editing the selection part to see if choosing a different area of the Persona helps?
It could be a number of different things, mostly related to genre of music, that could cause it. (Plus it's not a perfect voice clone system)
But this would be the first thing I would try to change because it's easy and not permanent.

[hollow jay]

I’ll check it out

[frozen dagger]

If you want to try to adjust your Persona (Voice, ugh that's gonna take some getting used to) selection..
I reccommend opening Suno in a new window tab, go to library, click Voice at the top to show only your Voices, then find the voice you want to edit, keep that page open in it's own window tab. This way you can just click back over to adjust again.
1 window tab for the Voice (Persona) to adjust, and the other window is where you are working to create music.

[thorny shuttle]

a dozen try's it still drops the intro in the first main breakdown after 20 seconds, trying to get it to be at least double this for mixing in on intro and mixing out on outro

man the sound on 5.5 is wild !!! so good

[frozen dagger]

I'm diggin the custom model a lot rn.

[acoustic glen]

RIP My Persona

Since v5 came out of beta and v5.5 was released, my process for making vocal groups is broken. I used to make a girl group/vocal group using the legacy model persona because it truly represented a group vocal with various vocal timbres. Now everything is broken, especially when using the cover feature; it generates a dull sound, the vocal timbre is just random, and it messes up the original version too much. For information, my workflow goes like this:

Extend/sample my loop, then make a demo using a non-persona.

Cover it using a persona (legacy model).

The result (v5 beta) persona legacy model:

I got the result that I wanted.

The result (v5 stable and v5.5 beta) persona legacy model:

Too far messed up the demo (original).

Sounds dull, low quality, and muddy.

I tried using the new persona voice model to get the sound quality I desired, but here are some catches:

As a vocal group, it only captures one dominant voice throughout the song; the rest of the group vocals are buried in harmonization (like backup vocals and ad-libs).

The cover goes too similar to the original (even with different prompts and a low-percentage audio influence slider).

For reference:
Here are the two songs I generated using both the new voice model and the legacy model:

Original (demo): https://suno.com/s/FCr6Xgt1mJvt1gLx

Legacy model:
https://suno.com/s/Rs7kfD4ik3DetL2P

New voice model: https://suno.com/s/Gs5dXdYAvweXEjlK

I like the vocal voices in the legacy model, but it sounds dull and messes up the original (demo). I like the sound quality and song similarity in the new voice model, but it doesn't really represent the vocal group I desired like the legacy model did.

Switching to v5.5, I still get similar results.

nb: persona i created from v4.0

[hollow jay]

What’s your feedback on that, what does it do specifically

[lone nova]

after v5.5 how do I create consistent artificial voice?

[hollow jay]

Generate without persona then right click > create > voice

[frozen dagger]

A simple way to think of it is like Insporation to the max. I think I put over a dozen. Not using any style prompts right now. The difference between the v5.5 gens and my custom is ridiculously present. lol.

[woven jetty]

what does this mean???

[hollow jay]

Can you show 5.5 and custom to compare

[hollow jay]

Retry

[woven jetty]

I tried using my uploaded audios for inspiration and also my voice to generate but it's not working

[hollow jay]

Early days system is 50/50

High demand too I think

[woven jetty]

what does that mean???

[frozen dagger]

Seems you can not use Voice with Inspo? I don't recall if I did that before or not. I know Voice (Persona) was incompatible with some features.

[frozen dagger]

i get the same error with inspo and voice

[unborn hound]

It keeps rejecting my voice. Is there any fix for this?

[cyan jewel]

Uh, this new voice verification thing literally kills the persona feature.

[frozen dagger]

have you tried multiple methods? (no background noise)

[frozen dagger]

voice verification only applies to uploads, not suno generated.

[cyan jewel]

incorrect.

[frozen dagger]

unless you're uploading suno gens?

[cyan jewel]

this isnt an upload at all.

this is a regular suno gen

not even a reuploaded suno gen which i do occasionally do if i took it off site to touch up in fl studio or smth

[frozen dagger]

ok, is it a suno gen that was made with a persona?

[cyan jewel]

its a regular suno gen.

not made with a persona

a regular suno gen.

no persona, no upload, made entirely within suno.

and it's making me voice verify

[unborn hound]

It makes you read out loud so not mmany options. I have a studio mic and a quiet environment.

[frozen dagger]

if we use the previous method to create a persona, to crete a voice, by selecting the track, there shouldnt be a prompt for voice verification. (unless there was an upload involved. upload or any derivatives.)

[frozen dagger]

it was having issues earlier. most seemed to be resolved. i was having a hard time when i was using a live recording. i opted instead to upload the recording to suno library and do it that way. when i did it this way i didn't have any issues really. (i think i repeated the verification phrase once, but music as on in background)

[desert imp]

yes i noticed that too...i guess just have to wait til the new version gets better :/ ive tried upload 24 songs still not as good as 1 single persona from legacy XD

[cyan jewel]

So. Genned a song without a persona. Try to create a persona of it. I get this screen. Something seems to be amiss with persona creation now.

[lone nova]

I have the same problem.

[wintry jacinth]

Was just about to ask about this

[frozen dagger]

what browser are you all using? same one?

[acoustic glen]

update
i tried making a new persona/voice from v5.5, and did a cover, the result is the voice sounds like original(demo without persona), not following the new voice/persona i made
sadly, all these new stuff really break my way of music making 😭😭😭

[desert imp]

me too, I just use legacy for now until the new model gets better

[wintry jacinth]

chrome

[unborn hound]

I got it working and GOD DAMN. It sounds like me singing

[pure panther]

I would try adding a "break" right after too. Sometimes you have to give Suno more than a nudge

[thorny shuttle]

I tried more lines didnt seem to adjust, what im trying to do is just generate a simple base line and kick drum for the 30-45 second intro and outro

Software always wants to jump straight into the first breakdown

[polar spade]

A new fresh song? We still cant use a song which has an upload in its chain at any point, as I recall.

[cyan jewel]

Suno-generated. Not an upload. Not a reupload after touching up off-site. A native suno generated song without a Persona.

[polar spade]

Got it. yea i see others are mentioning having similar experience.

[quiet solar]

Just to confirm, you're creating the persona through Library -> Select song - > ... -> Create-> Voice?

Sorry, still calling them personas from force of habit. xD

[cyan jewel]

No. I'm doing Voice > Create New > Select From Library > Select the song. This is what I'd normally do prior to the update.

Before it was Persona > Create New > Select From Library > Select Song

[quiet solar]

Yeah, okay...that's a UI issue. Doing it that way brings you into the new verification path.

If you do it through Library from the song, it works fine

[cyan jewel]

I noticed right as you mentioned lol.

[quiet solar]

It's probably been flagged up already, but I'll pass it along.

[frozen dagger]

@quiet solar a filter for uploads in the Voice UI would be nice too. When choosing Select from library. We have it as a filter in library but it is missing from this one. Where it would make a lot of sense. (Voice are uploads)

[quiet solar]

Good shout. Would be worth dropping that as a suggestion in https://discord.com/channels/1069381916492562582/1486542611475992749 if you haven't already.

[high snow]

Does anyone here have any ideas when it comes to making instrumentals to make them longer than 30 seconds or between one minute and two minutes? Because I wanna make an instrumental that’s like five minutes long or something but I don’t know what I’m doing wrong. Like I have my styles box and what I want for like instruments, etc. and the sound but it keeps like putting it either between 30 seconds or one minute. Maybe two minutes if I’m lucky

[young geode]

Yes.
https://suno.com/playlist/f3380023-03fa-4630-a7e0-3f4342dd7f39

You can use a combination of unspoken languages (such as Braille or Morse-code), in conjunction with the normal sections and heavy metataging.

[high snow]

The song link that you sent me of those examples that I can use in the lyrics section?

Because I did make one instrumental I have like a few lyrics that kind of repeat within the song but it’s only one minute and 58 seconds. I can send you the link if you wanna listen in.

[high snow]

Also, how does the Morse corridor braille extend it to make it longer? And does that mess up how the song sounds?

[young geode]

Yes.

And if you want to send it, go ahead, I may be able to help you with it.

[high snow]

Because the one that was talking about that’s one minute and 58 seconds I like how it sounds I just wish it was longer. I mean, I could use the extend future but sometimes it’s a nightmare. But I can always mash it up with another one.

[young geode]

No, though the it does effect the song's sound (in a good way).

SUNO does read both, because it uses an LLM as a translator, so it sees the language, it just doesn't speak the language, so it ends up extending the sections.

You can just take one of your songs, convert it into either/or, then insert it and it'll extend it by just doing that.

I have a feeling it also knows the words as well, so the emotional undertone does seem to come through.

[young geode]

Extention is a nightmare and I'd advise against it unless you have a song you already like that's either at it's generation end point or works up to a point.

[high snow]

Yeah, that’s why I might just make/generate a new song with your idea you gave me and then mash it up with the 1:58 sec song

[young geode]

Just one thing: I normally don't do any vocals at all in my instrumentals, because SUNO has a tendency to want to sing.

So if you don't want vocals at all during instrumentals, you should Gen that first, then use the editor to insert the vocalize sections

[high snow]

Well, I was thinking of just doing a few vocals which I sent to you. My one minute and 58 seconds cause I’m doing both instrumentals between having a little lyrics and none at all. But the problem here for me is just extending it.

[young geode]

Well the method I gave you will extend them.

The other thing you can do, is simply layer in the [Solo] and [Instrumental] tags, with many other instrument Metatags, with the Morse code sandwiched.

Solo and Instrumental are treated as sectional tags, like verse and chorus, and Vocals tend not to play during solos or instrumentals.

I was under the impression you wanted a fully genned instrumental, which is what my method was developed for, however if you're going that route, you'd should instead format it as such:

[Instrumental]
[Verse]
[Solo]
[Chorus]
[Instrumental]
[Solo]
[Verse]
[Instrumental]
[Chorus]
[Coda]
[Outro]```

[frozen dagger]

try duplicating your intro in the editor? cover/recreate. There is option to add new sections in editor too, results vary tho. But yeah, duplicate the intro, cover the intrumental to keep the intro vibe and then add the lyrics after.

[young geode]

If you need help developing the tags further, let me know.

[high snow]

Also, has anyone ever tested how many voices Suno can layer at a time when it is generating songs?

[high snow]

Alr

[woven jetty]

okay well now suno v5.5 is not even getting my voice right and I make sure no echo or reverb is in my voice memos

it's not mimicking it no more

[young geode]

Depends on what it is you want.

If you just want two different vocalists, for 5.0 it's as easy as tagging under each section [Vocalist: Sex, Range, Qualities]

So:

[Vocalist: Male, Baritone, Radio]

[Chorus]
[Vocalist: Female, Mezzo, Vocal Fry]```

Then in style just have ```<Vocalists: [Male, Baritone, Radio]; [Female, Mezzo, Vocal Fry].>```

That should work fairly consistently. 

But if you want something more complex within the sections itself, you'd have to do a lot of work on texturing the lyrics, and so far I've only been able to regularly introduce a Male, a Female and a Choir into a song all at once, with a 60/40 toss up on controlling who speaks by sandwiching lyrics between `[Male]`, `[Female]`or `[Choir]`, and having this many vocalists can cause SUNO to hallucinate.

[high snow]

Then go to #🪲┃bugs-web and report it there

[unborn hound]

What are putting you Style slider at to try and keep it close to your voice?

Audio Influence, I mean

[thorny shuttle]

Hey mate im having troubles forcing my trance tracks to not bridge into the first breakdown within 20 seconds, i want to generate a flat kick drum and base line for 30-45 seconds. Do you know any prompts to solve this issue

[young geode]

The only thing I can say is take out everything genre wise from your Style, insert only <Instruments: [Electric Bass Guitar: Model, scale, play methods] ; [Drums: Drum Kit Model, Rhythmic Kick].>, along side whatever other description you want to reinforce this.

[Instrument: Electric Bass Guitar, Baseline, Scale, Finger Picked]
[Instrument: Drums, Drum Kit Model, Rhythmic kick]

[Verse]
[Instrument: Electric Bass Guitar, Baseline, Scale, Finger Picked]
[Instrument: Drums, Drum Kit Model, Rhythmic kick]

[Chorus]
[Instrument: Electric Bass Guitar, Baseline, Scale, Finger Picked]
[Instrument: Drums, Drum Kit Model, Rhythmic kick]```

Take out effectively every instrument in the negative section.

Gen.

Then from the Gen that makes it work, extend it after 30s or so, add in your style prompt, get the whole song.

Then create a persona. And try and use that with heavy Metatags.

[thorny shuttle]

wow this is very different from how i approach generation.

[wild charm]

Is there a way to tell Suno to use your voice persona for specific verses or choruses?

[quiet solar]

I'd echo MasterofSFL's finding. Two distinct vocalists, with the hallucinated memory of a third being possible via Cover/Mashup, or the adaptation of one of the singers into group vocals.

But if you're trying to work with that and specify who sings which part, you're very likely to find yourself editing the song anyway. And at that point, you can realistically include as many different singers as your patience allows, if you generate them separately and don't require them to do complex harmonies. And if you sing harmony lines yourself as an audio guide, then the sky's the limit as to what you can do.

[quiet solar]

No, not in a straightforward way through prompting.

It's not possible to direct Suno as to where or how it should use something given to it as an audio input (e.g. Persona, Sample, +Audio).

[split jewel]

Hello, I've tried the voice feature a few times. It is very unconvincing about the quality and accuracy of my unrecognizable voice. You too?

[young geode]

How'd it go?

[thorny shuttle]

coudlnt get it to generate the simple intro

[young geode]

https://suno.com/s/2Tw8D9OGspgi1zgs

[thorny shuttle]

ill keep pushing it, very different style to my neiche genre im going for, but i can see the patterns.

i tried stripping my Style back to minimum input and i realised ive got way to much information for the output requirement, probably need more specfic prompts

[young geode]

Well what you're doing is trying to get an intro pattern going for the person/cover.

That's why you Gen the beat first without any inputs, then extend with the genre you want.

As long as the base is in line with what you want, further generations would take care of any specific sounds.

[neat jungle]

HELLO HELP )))

How do I separate the voices?

I use keys... example
[vocal: Woman] I love you
[vocal: Man] I love you too
[vocal: Woman & Man] We love around the world

The role can shift or break each time.

How do I properly sing with the correct roles?

[young geode]

https://suno.com/s/N34ZYtDtrYgXn3Z1

I used the same technique for this Western/Sludge Mashup

@thorny shuttle

[thorny shuttle]

i sent you a priv message.

[young geode]

Doing vocals line by line requires a lot of editing to have consistently.

However, if you're OK with them being added in certain sections or don't mind them flipping, the most basic metatag is:

[Vocalist: Sex, Range, Quality]

For Example:
[Vocalist: Male, Baritone, Chesty]

And you'd write your tags for each section like:

[Vocalist: Male, Baritone, Chesty]
Lyrics


[Verse]
[Vocalist: Female, Mezzo, Vocal Fry]
Lyrics```

Or

```[Verse]
[Vocalists: Male, Baritone, Chesty | Female, Mezzo, Vocal Fry] 
Lyrics```

[neat jungle]

Thnk ❤️

[pallid leaf]

I have studio sub, but can't seem to find the extract stems feature any longer. Maybe I'm blind but could someone help with this
ah nevermind it's there, had to set the slider to Clip

[young geode]

@polar spade I did a regen on the last section of this song, to strip off a very lengthy and hallucinating ending, and it gave me a major EQ boost that throws off the end.

Is there a way I can reduce that EQ in the editors or do I have to just try my hand at the regen?

[lone nova]

left click on your section

[young geode]

I knew it was somewhere, thanks mate

[torpid needle]

Has anyone had any success using their voice model with covers? Can't seem to get it to use my voice. It works fairly well with new tracks, just not covers

[unborn hound]

Anyone figured out how to remove audience sounds from a track?

[unborn hound]

It seems new tracks are like 80% my voice, and covers are like 50%.

[torpid needle]

If there's anything in the lyric tags or style that says "live" it will put in an audience.

[worthy kraken]

So for everyone who is suffering from extreme noise pollution (white noise, background screech, etc) that only intensifies as the song goes longer, i found a solution... sadly, not ideal and that will cost you 50 credits. But better than nothing.

You need to get stems (full) of your song, and then save them in some editor, and either remove or severly cut the "synth" stem. That thing in unbelievable bugged and creates and apart from the sections where it is needed, it creates a lot of noise out of nowhere.

Just be careful since part of it still can be needed sounds. But you can cut the noise out and leave those be.

[simple briar]

Yeah that actually makes sense. I’ve noticed that synth stem can get messy and add random noise over time. Good workaround tbh, even if it costs credits.
You could also try just lowering the synth instead of removing it completely so you keep the good parts without the noise.

[plush stump]

Hi there, does anyone know how we can create the old Persona’s still? I.e using a previously generated song for vocals?

I can’t seem to do that without ‘verifying my voice’

[sleek stump]

I believe you still can, in the create voice menu..

[polar spade]

I set my library list oldest to newest, went to page 70.
Click Create, and Make Voices was enabled.
Clicked Make Voices
It loaded the song sample.
I clicked Create.
It said created.
I checked Voices page and found my new persona created.

I then tried on a brand new song created by opening Create and adding a couple of prompts-- made the song. Then made the voice/persona, and it worked fine. No verification.

So it seems to work. It may be really picky on the file/song were using if it has any history in the chain of being part of any bits that may be greying out the Make Voices tool -- if its doing that.

But if you are getting a voice verification, thats always been on an upload or anything involving an upload in the chain ,as far as I know.

[sleek stump]

If you want to create a persona like from v5, I suggest to split and get a very clean vocal section thats about 1 minute of the best parts of that vocal, then use that cut stem for the Vioce.

[polar spade]

@plush stump That way ( above screenshot) seems to mean "your literal voice" or at least thats what Suno thinks. . And it will ask a voice verification of THAT voice thats in the recording, even if its a suno voice.

I think that, then, is buggy, and not quite the intent. Some mention of that yesterday.

I just tried to make a voice/persona using this method of that song mentioned above that JUST made a persona of. This way, it asked for voice verification and blocked me from doing what i just did.

But if we do the old way. Should not be a problem on non-uploaded songs or songs not persona'ed previously.

[sleek stump]

Ah, yep, thats likely a bug.. Need to get that reported.

[plush stump]

Awesome, thanks guys

[neat nexus]

What browser are you using? I was unable to access the option in the ... menu when I tried in Waterfox (Firefox Fork) and in Brave (Chrome fork) but in the official Chrome browser it was in the expected place with the name Voice and worked as expected. That was Version 146.0.7680.164 (Official Build) (64-bit).

I have passed this up the chain and they're looking into it.

[fluid vault]

Hello all, sorry if this has been answered already.
In the new suno 5.5 video, he talks about your voice sounding more like you, if you use the Audio Slider.

Is this the style influencer slide bar?

Am using chrome on android.

It also appears the App does not have the add my voice function?

[shy wasp]

What type of files does Suno want to upload vocal samples to build a voice? I have MP3 samples, but Suno won’t recognize them.

[fluid vault]

Did this work?

[polar spade]

Theres been some new tools added. Have you tried the newest Voice (persona) . You really shouldnt need to use the slider. It should emulate our voices fairly well.

But to answer, Audio Slider influences the strength of the Voice (emulator) applied to a cover or new song.

Voices may only work with 4.5 and above. To be honest, I havent checked in awhile. But your voice is not controled by the model, its laid into the output by using the Voice (persona) to copy ones Voice.

Now, having said that, are you simply singing on the mic or uploading a song that you sing on? Then, again, if thats all youre doing, then it is still the Audio slider that says you want to emulate the output -- i guess if youre covering your song you just uploaded -- to sound as close to the upload as possible.

[fluid vault]

i have not really used personas..

im just tryin to make it sound like me.. ive tried 4 different times, with voices.. and it just dont sound like me, so i dunno.. ;\

or is it the audio influence bar?

[polar spade]

Yes. Its the audio bar... slide it up to 85% or slightly more even on just an uploaded song of you singing to some music, perhaps.

If the lyrics match what is sung in the audio then you dont need the style slider up much anyway, but it should not affect the Voice from sounding the same as you have recorded , along with some pitch and quality enhancements.

[tacit valve]

So, can we change the vocals in a Suno song we’ve already made? Without changing the beat or the vocal style?

[fluid vault]

Ya i am unsure. I record my voice with my phone, and my style is like country outlaw ish. but it dont sound like me, ive tried 80% upto 100%. Sorry if i am not getting it.. ;\

[sour torrent]

@fluid vault Are you sure it doesn't sound like you? Does it not resembled what you recoded at all? Maybe you need to capture more vocals - how long was the sample you uploaded?

[fluid vault]

when i recorded myself singing i sang Jonnys daddy by Tim, should i sing the lyrics i want to upload? just thought about this, sorry. i sang for about half the time.

[polar spade]

Technically, yes, thats always been the idea.

Im not sure what you are doing, but if I take song sun by a female Suno vocal, I should be able to Cover the song, maybe set Male in the Advance area, add a Voice Persona of my own voice.....

Slide Audio anywhere from 25% to 89%
Set style to whatever fits my style and genre tastes

And it should then sing the same song having switched my voice, assuming some variation in the melody and some progression of the song from the original.

[tacit valve]

Yes, I can produce different variations, but I was wondering if I could get exactly the same result. Thank you for your help.

[polar spade]

Can we match the melody note for note though? probably not.

Thats still defined as "cloning" and that likely still takes a pass through of channels to take the original, covert to MIDI on the backend and then back to a new vocal (instrument) which other apps may do. Its not set that way with Suno as of yet.

[tacit valve]

I see

[polar spade]

Yea. You said "beat and vibe".... and those two often remain, depending on how Suno reads the sliders. We may have to fidget with combinations to find a good setting.

[tacit valve]

There are loads of new features. I hope I can replicate it to a large extent, if not 100 per cent. After all, it just needs the right settings and code.

[polar spade]

I take a song from one of my duo female groups and then have my country band sing the same thing. Never had much problem with thr melody being nearly identical.

[sour torrent]

@fluid vaultI think that might help - but I think people are seeing differing results. The style is important, the output will depend on the input greatly. If your (pitch range) input is narrow and matches the style of the output wanted it should sound the same. The broader the deviations the greater the chance of it sounding unlike theb source I believe.

[fluid vault]

ill just keep trying with this info. Thank yall very much!

also, Chrome on Android dont have the Audio slider??

[sour torrent]

@fluid vaultAre you certain you have your created Voice selected? (just checking).

[fluid vault]

yes?

[polar spade]

Keep in mind, now with Custom Models, we could almost eradicate the need for Voice for new songs or even Covers.

Upload 20 songs or more (the more the better I understand) that are just your voice (and whatever musical styles youre singing) ... or... a favourite AI band youve been creating. Let it do its magic and try it out on a a new song or a cover.

Its concept is that it will infuse your style as well as sing with the voice it trains on when using that Custom Model on your creation.

[sour torrent]

@fluid vaultYes Dark yo uare correct - I just checked my Android, there are no voices option.

[fluid vault]

ahh okay

pushes desktop version

NOO!

still dont work

[tacit valve]

So these new features can do even more than I thought. Thank you very much, I’ll give them a go straight away.

[fluid vault]

well i will keep trying via phone, and get to maybe a less windy area to record. i was outside, low wind, but phone can pick up quite a bit. and sing the lyrics. Thanks again @sour torrent

maybe i wont have to touch the audio bar

[vague scaffold]

is the suno v5.5 bug fixed yet

like the extend bug

[tacit valve]

Instead of speaking directly into the phone, if you speak into the wired headset microphone, it will pick up the sound properly. That’s what I do.

[polar spade]

Its new to me too, so youll just have to test along with the rest of us. But ive made two models and was impressed thst it finally got a soecific vocal I had back in 3.5 model that I could NOT use in covers as a persona since 4.5 came out. So, I might be able to bring that band out of early retirement.

[tacit valve]

To be honest, this update has made me very happy too. I was working on my concept album with Suno, and now I’ll be able to tackle more varied and complex projects

[lost forge]

@polar spade sorry for pinging you directly, but you have been here as long as myself and as a professional lurker I recognize a knowledgable and interested person when I see one. I've been struggling to get a good quality out of 5.5, with the sounds being tinny and it feels like the shimmering of 4.x time makes a return. Is it just me or what is your perception of the new model?

[shy wasp]

Anybody know?

[tacit valve]

did u try wav?

[polar spade]

The other issue was consistency. So when making a custom models, be very specific on NOT having any other sounding vocal in the uploaded files. Im going to make more segmented samples of 30 second- 1 minutes snippets of just the specific vocal as part of my 30 to 100 WAV files, in hopes that locks in the vocal quality of just one vocalists.

But its theoretical, if we include three or four vocals in the mix, suno may include them all in one output, that would be cool if it did, but not exactly expected unless we can prompt fof it as well. We'll see.

[shy wasp]

Not yet. I was hoping to get some directive before I started wasting time converting files.

[polar spade]

Its WAV.

[sour torrent]

MP3 also works fine for me (on Windows)

[tacit valve]

This is just the first update; I think more features will be added soon because I’ve heard they’re doing a great job. They’ll be working on it.

[polar spade]

It may be the genre as well. I get artifacts and a hiss or shimmer here and there but not across the song. But ive made less orchestral and distorted songs recently as well, so I may not be the best to say , in general, how its affecting all genres.

Of course other comments have been popping up saying a shimmer is present. I just dont knownhow much is anecdotal, how much can actually be show on a spectograph, and how much is just Suno not quite getting the synths right and keeps including a shimmer pad by mistake.

[polar spade]

And yes. Ive seen you pop in last year, as I recall -- Octoberish. Good to see you my friend.

[vague scaffold]

still waiting for my answer

[sour torrent]

@vague scaffoldDoesn't work for me so... guess not.

[lost forge]

Thanks for the reply, appreciated. Next on the testing table, custom models. Sounds interesting.

[tame verge]

Not sure if this is the correct thread for this, but does anyone know how to make male and female voices sing on their correct lines? Or, if there’s a way to cut and paste sections from different versions of the song in order to combine them?

[tepid moth]

I tell yah, some of the pieces of music that I listened to within the last 24 hours made with V5.5 I thought were astounding. I do think that there's a bit of fidelity boost for sure. I did like that there was a natural progression in the songs in regards to building energy. I get very disapointed when using the cover feature and my second versus run completely off the rails. I end up having to tame the second versus down in "Studio" just to make them useable. It may be that I'm using the "Weirdness" slider correctly. I may have to back below 25 or even lower. A great feature would be if we had a "weirdness" slider for just vocals. Anyways, as always, I do enjoy playing with Suno. I just wish all the copyright and ownership usage could me a little bit more definitive. Way too much speculation floating around on YouTube.

[golden osprey]

Any tips to get a deep male voice? All of the usual things that used to work don't work at all right now, no matter what I'm putting its like suno is incapable of giving me a deep male voice unless I use baritone and then it's like barry white which is just too much.

[hard beacon]

what kind of music do you want to create?

[tepid moth]

Slow a alto down, change the key, stem out the vocal register you like, and use it within the remix/inspo feature.

[golden osprey]

some kind of folk rock ballad, I'm starting to think its the emotional part that's overwriting the voice direction

which is a bit annoying

[hard beacon]

ok.. well.. can you give me reference well known artist for that style? 🙂

[golden osprey]

I have no idea

[hard beacon]

stlyle prompt example: Folk Rock Ballad, Slow Tempo, Melancholic, Warm, Acoustic Guitar, Fingerpicking, Soft Drums, Bass Guitar, Male Vocals, Deep Baritone, Gravelly, Raspy, Smoky, Emotional

[golden osprey]

I'm just making this for fun, I'm not muscially literate at all, I just put whatever Grok is suggesting me, I've made hundred of songs like this no problem but now its not working anymore, which is funny because before it was impossible to get a normal male voice, everything was deep lol

I did that, and everything similar

nothing works

[hard beacon]

and in the lyric prompt you can use [Verse | male vocal | deep baritone | gravelly | raspy | smoky] as metatag (is metatag the correct word??) 🙂

[golden osprey]

yeah tried that too

[hard beacon]

if you want you can send me the lyrics via DM and i can try to create some outputs with different prompts. Maybe there is a working one then for you 🙂

[golden osprey]

husky low gravelly male voice also gives me some 20yo male voice, its hilarious when I think about it but also frustrating to waste credit

its really bizarre everything was fine a week ago and it didn't suddenly happen yesterday with 5.5 either it started at the beginning of the week I would say

[peak lava]

I think 5.5 it's a downgrade of 5. When I use the exactly BPMs, SUNO makes anything diferent, sometimes much faster. And with the same music style, makes totally different styles, crazy...i feel i'm wasting thounsands of credits. v5 its ok to work, but now...

[hard beacon]

@peak lava : the thing is, that we all have to learn a new modell. It's the same "problem" with every new modell, but if v5 works fine for, then perfect 😉

[vague scaffold]

well okay

lets just see if they fix it or not

[golden osprey]

I just tried to do one in v5 pro instead of v5.5 and it seems to work better there at least so I guess its really v5.5 thats bugged after all, must mean it's been cooking behind the scenes all week. I'll stick to v5 for now then and see if I can get what I want that way

[hard beacon]

like i said.. if you want i can try to create some outputs for you 😉 just an offer

[golden osprey]

thank you for the offer, thats really nice of you, but I'd rather it works for myself 🙂

[hard beacon]

no prob 😉

btw.. if you use chatgtp for some help i would recommend the SUNO GPT by ChillPanic

[viscid ravine]

You misspeld chatgpt

[hard beacon]

oops 😄 😄

plz 4give me 🙂

[shell loom]

misspled? misspelt 🙂

[hard beacon]

hahahaha

PWND

😛

[ripe galleon]

will suno recognize my (awful) trumpet playing as a trumpet?

or something else

im genuinely bad at the trumpet so keep that in mind

[hard beacon]

hahahahahahaha.. made my day

[ripe galleon]

you are welcome

[hard beacon]

try to add it as inspo for a dubstep song 😄

[ripe galleon]

i could use it as a bass lol