📚┃suno-school

[neat nexus]

The model has a mind of its own sometimes. That's why it's artifical intelligence and not just a word processor for audio.

What I do when this happens is check my punctuation for anything that might be directing the flow of vocals. Are there comas? Try getting rid of them. Don't have any? Try adding them.

Does your lyric flow as one piece or do you break it up into its own lines? This can have an impact.

Trying different things will give you ideas on what works best for you. You may also investigate your wording to see if your lyrics just don't flow well from subject to subject. The AI may be choking on something that doesn't sing well.

Are you resuming from an earlier part? Try resuming from a different time, a little earlier or later. Give it some different music to work with.

Bottom line? You're going to have to adjust and try again and adjust and try again. Maybe investigating the style prompt will help? Maybe you need some instruments between parts to improve the flow of sound? Going to have to try things until you get something you're happy with.

Good luck to you!

[spice furnace]

Well, im fighting with it currently, and its not my first rodeo, but this issue always comes up in some form 😦

i tried braking the phases in different ways, adding/removing commas, merging th e whole block into one, nothing seems to work

i jumped here so someone might found the solution for this 😦

[violet tundra]

The reason i've found is that 1) my custom lyrics do not flow, and 2) the extend time needs to be adjusted a second or two in one direction or another. This should fix it.

[graceful condor]

Can you get suno to do a duet? [Bridge - Male and Female] [Breakdown - Male] I've tried stuff like this but it seems to not work

[winter bay]

I did hear a generated song that had a duet, but the user never specified it to be a duet, so maybe it happens on random occasions

[hollow hare]

Basically it is currently hard to achieve. I had some luck with swapping male and female verses but duet Chorus is still a coincidence

From my tests, I found you are more likely to achieve the duets if you do not specify vocals in the style prompts (as male likely has low weighting, it is typically overwritten) but you can specify the gender in the lyrics as you did there.

Here is my test playlist with various methods I tested:
https://suno.com/playlist/11ff3827-bdcf-4ef5-bd04-da8647c6cdb6

The only ones with the duet choruses I can recall from the top of my head are in Love in the Moonlight, Eid Celebration (at 1 min, completely by chance), and Harmoney in Tune 1.2

[graceful condor]

Thanks!

[dapper crane]

can yall say generated please, have some respect for the machine

[unique stone]

Maybe they'll tighten this up when there's more control over what other users can do with your stuff. As things are, if I share a Suno link anywhere I assume that it could be extended/taken/whatever so I just keep private anything I care about.

[cunning mulch]

have to wonder if those IDs are useful for anything in song generation

[naive quail]

I try to make of for their lack of extension use...

[delicate nebula]

Guys, can you tell me the best way to extend some song only for 1 or 3 sec no more than that? Thank you.

[brittle apex]

It was live briefly yesterday. I used it, it was nice

https://suno.com/song/c997f590-e993-4a97-80d1-ecad73b1e8b4

lol, it helped me get an ending with V3.5, without having to switch to V3. V3.5 always wants to make some piano piece instead of ending

[unique stone]

My theory is that if you give it nonsense in the prompt that Suno can't figure out and then put "piano" in the negative prompt, Suno will just implode.

[naive quail]

starts to drool

[brittle apex]

It literally didn't know what to do so it gave me blissful nothingness and a credit refund 😉

[unique stone]

There seems to be a few things bubbling away under the surface that aren't quite fully-cooked yet. Interesting to see brief glimpses of them. Pretty sure I saw something in the data related to inpainting as well but that's known to be coming with v4 so no surprise there.

[naive quail]

This is how Sunonet becomes self-aware and activates the launch codes.

[proud spruce]

Can suno use 2 different singing voices in 1 track?

[lofty merlin]

Am I being dumb and/or blind here? Where is the "Edit Displayed Lyrics" option? I want to remove some duplicated lyrics from my video that resulted from the first generation getting cut off at max time, but I can't find that option in any menu, despite the changelog having a whole thing about it...

[naive quail]

Scroll down to the bottom of your lyrics

[lofty merlin]

There's nothing there. The lyrics end and then it's just the bottom of the page.

[naive quail]

Doh! Yours is awol!

[lofty merlin]

O_O Apparently!

Looks the same for me in both Firefox and Chrome...

Well, that makes me sad 😦 But I'll put it on YouTube anyway and just complain about the confusion in the video description 😅

[night jewel]

Good distributor you know which could help me in getting my first song released within a week,
I tried jumpster but have an issue uploading my audio
I wouldn't want to pay for yearly for yearly subscription, right now may be going to release 1 or 2 this whole year

[shut moat]

how would I go about making breathing sounds in the lyrics.

like a loud sigh

[marble wadi]

idk what changelog u r talking about lol

[lofty merlin]

The "What's New?" pop-up

[marble wadi]

lol wut

whats that even do

[lofty merlin]

It shows a changelog of the new features, in reverse chronological order 😂

[marble wadi]

oh cool

i never read that

[lofty merlin]

Since Chirp was originally based on Bark, you can probably just do something like [Sigh] in the lyrics. I don't know how well it will work, given the amount of music training that probably doesn't have much representation for things like that anymore, but... maybe somewhere deep in the model, it still knows.

[shut moat]

i thought the brackets were song directions

[neat nexus]

You are a bus driver.

You drive down main street and pick up three people. At next stop five people get on and two people leave. You turn down to maple drive and pick up eight people, go to the next stop and two people get off and eleven get on.

What color is the bus driver's hair?

[marble wadi]

um my hair is brown

[neat nexus]

What time is it?

[marble wadi]

its 6:09 am

[shut moat]

11:05 am here.

[lofty merlin]

Not exclusively. It's just various things the model has learned from its training data. A lot of lyrics use song structure markers, so it learned those, but the prior Bark model was trained on closed captioning/subtitles, which uses the same square brackets for non-speech sounds. So Bark learned a lot of those, too.

[neat nexus]

How many letter 'R's are in the word strawberry?

[marble wadi]

um two i think

[neat nexus]

Tell me more.

[marble wadi]

nah thats it

[neat nexus]

Tell me more.

[marble wadi]

seriously theres only two

[shut moat]

I'll try it, though I don't think there is much control yet, just suggestion at this stage in the bot's development.

[hollow hare]

What's updog?

[marble wadi]

not much wut r u

[lofty merlin]

Most prompts into generative AI models are just suggestions, when you get down to it 😉

[shut moat]

I been using Midjourney since nearly the beginning, there is an incredible amount of control with that model.

granted I have WAY more experience using it.

[lofty merlin]

To be clear, a Midjourney "prompt" is a mix of settings and prompt.

The prompt itself is still just a suggestion 🙂

[shut moat]

image generators and music generators apples and...steaks.

no where near the same I imagine.

English has specificity where I lot of other languages have range in meaning in their terminology.

there's a whole slew of words and phrases that have specific interpretation in MJ>

especially Midjourney's mistakes that are mistaken specifically.

Example of such a specific mistake is a French heirloom scalloped pumpkin variety called "Patisson strie melange" in the V 5.0, 5a, 5.1, and 5.2 models Midjourney interprets this plant variety as a command to convert the image into fabrics.

Now this mistake can be directed and applied with image prompting to be utilized in product and clothing design, without needing to "suggest" to the bot of what kind of clothing you want to generate.

[lofty merlin]

That seems less like having "more control" and more like you've just learned the ways in which Midjourney has mislearned its embeddings 😂

[shut moat]

mistakes are valuable

especially if they are interpretted in only 1 way.

[lofty merlin]

Sure. But my point is that it's not like some kind of control built into the model, it's you exploiting a model failure. Which is fine, but not something you should expect from other models 🙂

[shut moat]

and some english words are that specific. like damnatio memoriae, apotheosis, nadir, ecuminopolis, words that have an extremely narrow band of interpretation.

I don't expect any other model to be as comprehensive as Midjourney.

they are at minimum 1-2 years behind.

the control part is the 50+ commands and the strength of each of those commands. There is no other model that offers that degree of control.

[lofty merlin]

You haven't used a local install of Stable Diffusion, have you? 😉

[shut moat]

I have used some SD, but my computer is not robust enough to do it on it's own.

[lofty merlin]

What kind of graphics card do you have? (And how much VRAM in it?)

[shut moat]

only 8GB Vram.

it's an old computer, saving for the GB200 in 2025.

[lofty merlin]

I also have 8GB and run a local install. The Automatic1111 WebUI (or the more optimized fork, Forge) has a command line flag for --medvram (or something similar; check its docs) that moves some calculations onto the CPU, letting you run full SDXL on only 8GB of VRAM 🙂

[shut moat]

now is not the time to be building a PC, especially with what's coming in just 1 year.

[lofty merlin]

I use Forge with SDXL, ControlNet, and Layer Diffusion. I love it.

[shut moat]

the 5090's and the GB200's and B200's

I'll be some of the money I made from NVDA's stock going through the roof to build it.

[lofty merlin]

Sounds great. I'm just saying, you don't need a fancier graphics card to run it now 🙂

It'll speed things up, and let you use bigger models, sure. But for SDXL + extensions? Not needed.

[shut moat]

I want to train a model on just my own artwork.

all my drawings and paintings.

[lofty merlin]

Well... alright, for training, you will need a better card 😂

[shut moat]

yeah

I have about 600 art pieces to upload.

and the macros of each of those pieces.

[lofty merlin]

Or you can rent a Replicate rack for a cheap amount to train the model, and then download the trained model to run inference on your own machine. Probably cheaper than buying a new GPU

[shut moat]

no everything I have done by hand.

[lofty merlin]

Suit yourself

[shut moat]

I want a "me" model.

[lofty merlin]

Right, I meant use the Replicate rack for hardware to train a model on your own work

[shut moat]

I'm not terribly technical in all of this, SDXL was hard enough to get into since I can barely understand coding.

I get the Comfy, I can understand it visually.

but writing...

[charred merlin]

ftr, you ABSOLUTELY can train a LoRA on an 8gb card, it'll just take longer 🙏

[lofty merlin]

Shouldn't need any coding. You just make an account on Replicate, and search for SDXL Dreambooth (if you want a custom SDXL model) or an SDXL LoRA training toolkit (LoRAs are custom models that can be applied on top of the main model). Most of those just ask you to pick the main model and then upload your images.

[shut moat]

I donna wanna wait.

[lofty merlin]

True... much longer 😄

[shut moat]

lol

[charred merlin]

comfyUI is a superpower because unless you strip the metadata from the png, you can literally drag and drop a comfyui result back into comfy and you automatically get the workflow and seed that generated the image 👀

[shut moat]

I tried making a permutation, that was 100 jobs but the resolution I wanted said it would have taken like 80 hours.

😐

[charred merlin]

Sounds more like the config wasn't set up properly, that's wild

[night jewel]

??

[lofty merlin]

Personally, I hate flowgramming... but that's because I've been coding since I was 12 mumbles years ago, so I'm an old curmudgeon who likes my commands not to take up chunks of my window real estate 😄

[shut moat]

the res i set it at was 6 x 4k

WAY too high.

[charred merlin]

If you only want a single song or two, I've heard others suggest a service called SoundDrop, but I haven't looked deely into their pricing (I think they take a 20% cut)

[shut moat]

but upscaling smoothes features unintentionally.

why not start high i say.

[charred merlin]

The only other one I'm really familiar with is LANDR, and I'd really recommend them, even if you just sign up for a month to distribute then cancel -- not recommending to just sign up for a month, but if you don't need all of the services and mastering they provide 🤷‍♂️

[night jewel]

Ok

[shut moat]

anywho, I gotta work on a commission. a cat, in pointillism, yay...

[charred merlin]

https://cdn.discordapp.com/attachments/1271789804677501039/1273848894299570186/20240816-0339-46.0509229.mp4?ex=66c01bbb&is=66beca3b&hm=142cd3f3474dca83731e16067988f317f8be1e0e83ba68e396c233a02c0b084e&

I hear you, and I'm usually 100% there with you, but this is just fun

Yes, I know it says Takking. I could either have fun with cloud-based Flux or play with with Comfy on my 4070ti.

...gonna go have fun with Flux again 😂

[lofty merlin]

Today at work, a teammate did tech support for someone using Nuke, which is a special FX software that uses flowgramming for workflows. After 30 minutes of troubleshooting, he found that the user had duplicated a node 10 times and hadn't connected the duplicates to the graph... but he couldn't tell because the nodes were all stacked on top of each other.

I'm sorry-not-sorry, but that kind of thing is literally impossible with code 😂

[leaden lark]

And then me with my 3060

[lofty merlin]

Me with my 3080 Ti... which is only marginally better 😅

[night jewel]

Finally done with jumpstr , Anyone have Some experience wrking with them ?

[gusty acorn]

Is there a better third party option to extract stems? Online tool or desktop application doesn't matter.

The stems option in Suno is still just not good enough yet.

[hollow hare]

I heard good things about BandLab, see #📚┃suno-school message

[unique stone]

UVR works well

[eager topaz]

RipX DAW Pro is about the best IMHO. They offer a 21 day free trial to assess it. It's all I use and it does an excellent job

[cunning mulch]

.... it's getting a bit dumb how it has a line ready for it to go, same meter and all, but it says "nah, hell with it, lemme just repeat stuff from before"

[cunning mulch]

like, seriously...

and the WORST part is it acts like it's ready to go to the next section, but then "nope, lemme just ignore what you've put in and go back"

[cunning mulch]

I swear, I'm already 4000 credits into this damned thing

[timid wasp]

in these moments I'm getting max angry and confused. Hugs.

Try to take a step back, sometimes a little tweak in a not so obvious place can move a mountain

[cunning mulch]

I wouldn't be so upset if it wasn't the same places

and almost always the same way

[true night]

4k you're creating something wonderous then

personaly i hope i never use 11k on 1 song again. its uneven numvber i want at least 2 songs pr 10k

[cunning mulch]

well, I just want it to come out close to what I'm looking for

and it's easy to go thousands in when every generation fails and repeats crap you don't want

or it just rushes through everything

or goes flat in tone, or...

[soft gyro]

one more month gone and nothing changed in the UI except the red dots....and there are still so many needed things missing ! what other programmers can do in a couple of days Suno cannot do in 4 months

[stark spear]

Hey, i have a question: are there any promts or is there any known syntaxt in order to let suno sing a specific word in a different Language? MotoGP - but he sings it in german - sounds awfull, he shoud sing in in english - the rest of the song should be in german, does anyone know how to?

[stiff sable]

i dont believe there is a specific way to outright tell suno what you're trying to do, but rather than writing the full word as normal, you could try spelling it out phonetically in the specific way you want it produced which might help get you in the right direction. 🙌 again, it might not work every time but you might get a bit more luck with it

[flint sparrow]

Guys I have a ready-made track, but it has a too Sudden start and I would like to add a small instrumental intro. How can I do this without breaking the whole track?
The task seems to be simple: add an intro. But I can't figure out how to implement it...

[naive quail]

Not currently an option, unfortunately.

[lethal pond]

Hello

[cunning mulch]

this is what I'm talking about with repeated lyrics... not only that, it just flat doesn't read things properly. Shame too, 'cause it was pretty decent: https://suno.com/song/ac76c804-7c2b-4a2b-a985-603b68e82b40

but, see what I'm saying with patchworks of 'flags' and hearts?

[naive quail]

I think suno's struggling with the lyrical flow for that one and essentially tripping over its own feet. Suno can be pretty fussy about that and will try to make songs conform to something that it thinks is right, come hell-or-high-water

[warped furnace]

If you want changes within a song it's best to cut the creating process into smaller pieces. And if you give too much direction in brackets most of the time it won't work. Instructions like [hook] (if there is one) and [interlude] have worked for me. Also changing the syllables count in verses and chorus (and interlude/hook).

[cunning mulch]

well, that was me tossing the original out and starting anew

that's why it says v3

[warped furnace]

I meant working in sections. Like starting with intro and verse or chorus depending on the structure of your song. Keep it going to the point where you want a change. From there you extend your song with fresh style or slightly altered style instructions. Maybe a different tempo or some such. I think you'll find extensive discussions on that subject if you search. Theharber has left many useful comments on the creation of songs. 🙂

[cunning mulch]

my 'ideal' version if we ever get the ability to edit lyrics in an entire song, is this one: https://suno.com/song/d844660d-0bf8-4951-b19c-fad7f6845a62

[runic osprey]

Do you guys think they will ever make an update , as you have the upload audio , to be able to like record you humming a melodic line and the ai to take over and try to make something along those lines?

[cunning mulch]

I know he has, but the whole 'just extend' answer is a workaround to a deeper issue

namely that 3.5 is just a bit too headstrong in trying to ad-lib early on, and enforcing structure it thinks is right a little too harshly

[naive quail]

Reworking the structure a bit will help, as will adding interludes or solos before the parts that Suno ends up repeating... seems to allow Suno to sort of catch is breath when it's struggling with the lyrical flow. Here's one with a single take: https://suno.com/song/93673469-af16-4680-974a-cdc18e38d65c

you're probably going to have to fight suno and use extensions to get all the double/half times

[cunning mulch]

hmmmm

fair

now, what would happen if I were to maybe structure it as: 2-line Intro, 4-line 'Verse', 2-line Pre-Chorus?

[naive quail]

It all just depends on how the lyrics flow... and the current phase of the moon. 😄

[cunning mulch]

it's kinda got a 2-line Pre-Chorus anyway, so I might as well...

[naive quail]

...and the "Singer" that Suno ends up using.

[cunning mulch]

but what gets me is how it properly terminates the end of the 'verse', almost like it's ready to switch timings, and then... just doesn't

and it also seems like the 'Intro' is more likely to use a quieter, almost effect-using voice

in that one I want to re-use the instrumental for, it feels like those first two lines are delivered as an intro

I'd also have to see if it can be coerced into keeping coherent if I feed it A,A,B,C,B,C,D,D rhyme schemes...

but probably not

it's fine. I kinda like the energy on lines 1-2 being low, bumping up in 3-6, then swelling in 7-8

my so-called 'Outro' might wind up being labeled Chorus 2 to try and get it to keep the same melody too

[naive quail]

note that using multiple [Chorus] tags without the lyrics after the first one are much more likely to result in hallucinations the longer the song goes on. I've had it work well on short tracks, but it always bites me in the arse on longer ones 😄

[cunning mulch]

hmmm

I did that to keep character count down

[naive quail]

Suno ends up with weird repeats or shuffles some of the lines

[cunning mulch]

it does

now, I notice you used [Chorus: (blahblah)] instead of each one being its own tag

does that make any difference in it being followed?

see, under the ultimate ideal situation, I'd hit Generate, and get essentially a carbon copy of the Public one, just with the lyrics I'm using now set into the meter

100% fixed, finalized and done.

[naive quail]

maybe? 🤷 I stuck with what I know 😄

most tag descriptors like that are wishful thinking on our part, but that doesn't stop me from using them

[cunning mulch]

if the universe wanted to throw a bone on top of that, it might give it a tiny bit better intro than just jumping right in...

at least I was able to use the instrumental stem to check BPM manually, and found that what seemed like a full 4 beats clocked in at essentially exactly 1 1/3 seconds, so it was easy to calculate...

[small sun]

i accidently released a track with the same title as a Stevie Wonder from 1981, chat gpt says i should probably take it down and rename to avoid any legal complications. i dont know, what do you thnk?

[vestal rapids]

there are a lot of tracks with the same name, don't worry about it

[graceful condor]

Hi there! Does anyone know when Suno 4.0 will be released? I'm looking forward to it.

[vestal rapids]

yes

[graceful condor]

When will it be?

[vestal rapids]

when it is ready

[maiden sparrow]

Would be cool if you got a song u like created, and then, just be able to change the genre version of it. I am not complaining tho

[cunning mulch]

I'd settle with just minor lyrics changes

[polar spade]

Titles are yet to be copyrightable, but certainly the closer it gets to iconic songs or titles , the more you'll be questioning your motive. I've yet to title a song " Hey Jude" and likely never will.

[polar spade]

But if it's a one word title , like "Superstition". Nawww, you good.

[polar spade]

But always good to be creative. Rename if it still works, is a respectable suggestion. Some, however rely on their similar title being searched and one searching happens to stumble on one's similar title, thus more chance it might be heard. Question your motive and decide from there.

[polar spade]

But just click " reuse prompt" (in your menu) to get the same lyrics , then change your prompts to get a new genre. That's the whole idea with suno. You can already change genres.

[polar spade]

Same melody and tune, just inpainted or even directly change words (or just typos). Yep. I'm for that.

[polar spade]

You can, at about 10% luck, change the vibe within a song, buy changing prompts on an extend. Usually, just use one word prompt to force it. Most times though, it's more of a change of nuances. A new instrument, a tempo change, a distict vibe change. Maybe a whole genre -- though I've not gotten that yet.

[cunning mulch]

I've had wild swings like that... not fun

well, you know you have one if you get a tree image when you never saw anything like that before

[hollow sapphire]

Hi everyone! I joined Suno with version 3, and then 3.5 came out. So, I was wondering why the "chirp" was removed and what its purpose was. Was it some form of direct programming? Yes, I know, I'm curious! But that's how we move forward! Lol😂

[hollow hare]

Chirp was a discord bot/interface running Suno v1. It was decided to be discontinued a few months ago as the majority (>99%) of people utilized the website with the latest databases instead of the bot.

If you want to know how the bot worked, there are several videos on YouTube demonstrating it

[hollow sapphire]

Ok! Thank's so much!!! 👌

[snow cedar]

It's terrible, I've generated songs so many times, but nothing like today has ever happened. I specify the required voice style (female), but the result is always verses with a male voice—80% of the generations are like this. Credits were wasted, and the result violates the style instructions. Please make it so that a specific voice can be selected, and the AI won't be able to ignore the given instructions.

[onyx glade]

Woah i didnt know you can guide an instrumental like this: ```[Intro]
[sad, scratchy violin solo]

[Instrumental]
[violin, cello, minor key]
[forceful tremolo strings flourish]
[Distorted synth melody, fade in]

[Instrumental]
[strings, synth, swell]
[Drums, kick drum, snare]```

Where do you put this? In the lyrics section?

..🤣 ofc

[naive quail]

With varying degrees of success... a lot of that stuff won't be recognized by Suno and will ultimately end up being ignored or used in unexpected ways.

[charred merlin]

No harm in asking ❤️

I find it helpful to keep in mind that Suno doesn't look at these like explicit "instructions", but suggestions that it's able to ignore or focus on. Knowing how the style prompts, metatags & lyrics work together is an evolving process -- using community resources like prompt lists, and seeing what others have used to make their songs is a PHENOMENAL way to learn and get ideas, but I find people have a lot more longterm success if they rely more heavily on their own experimentation.

Even if no one else understands what you're doing, as long as YOU do, that's the important part. @metalzealot describes Suno as a "vibe interpreter", which I absolutely love -- no right or wrong way to use Suno, just vibe with it and see what makes sense to you 😄

[jagged narwhal]

I generated a good song but I didnt like vocal so if i click extend can i get only music?

[robust ermine]

Is anyone else noticing that v3.5 gives very similar outputs? E.g. if you input "Russian dark pop", it will almost always give an Am - F - Dm - E harmony. Just tried, not the case with v3/v2

[charred merlin]

#📚┃suno-school message Very common

v3 excels at giving broad and versatile outputs, but v3.5 tends toward "samey" results, or it sounds like generic pop/rock/rap/dance, etc, unless you have a strong musical hook to continue from in the extension, or strong use of prompting.

[robust ermine]

lol. I think I'm gonna try starting generations with v3 or even v2 and continuing with v3.5 then

simple example, the prompt is just "nefarious"
v2: weird, as expected: https://suno.com/song/f024fe55-4060-48c0-b1a1-3a61d1b3a19f
v3: as well, although the second one was just piano: https://suno.com/song/3bb321f0-745a-4f4f-870c-e3db28dadedc
v3.5: always just piano, some generic "beautiful" melody: https://suno.com/song/afd099f2-b6a8-4891-b69c-52ecb12db013

and v2 sounds way clearer in the sense of instrument separation/clarity

v3.5 always gives piano with "nonsensical" prompts

[charred merlin]

#1256301881966198805 message Here's one of the Prompt Experiment Playlists from the pool party thread where I used @naive quail's rainstorm prompting to try to affect a lightning crack -- v2 stays rain, 1 of the v3 tracks swapped to music halfway through, and v3.5 swapped to admittedly amazingly blended metal in both instances.

[robust ermine]

hehe

[toxic minnow]

I have done a lot of those tests... my conclusion: the better results are achieved by making the song in very small parts, because giving too many instrumentations guide in one pass, is simply too many TOKENS for the SUNO MODEL to process in one shot... but one at a time verse/chorus/etc.. it is incredible what it can understand... UNTIL it thinks the [TAGS] as lyrics and sing them !!!

https://suno.com/song/c532e15b-0596-410b-91a9-e88b05998664

[Intro]
[Distorted feedback and heavy guitar riff]
[Crashing drums]

[Verse 1]
[Heavy, distorted guitars with palm-muted power chords]
[Pounding drums with emphasis on toms and snare]
[Raspy, powerful vocals]
Shattered dreams on a concrete floor,
Screams of anguish, we can't ignore.
A world consumed by greed and hate,
We'll break these chains, before it's too late.

[Chorus]
[Guitars with open chords and added reverb]
[Soaring vocals with a touch of echo]
[Layered background vocals for anthemic effect]
Broken crown, cast it down,
We rise from ashes, a defiant sound.
Scorned and bruised, but never tamed,
Our spirits burn, forever inflamed.

etc...

[toxic minnow]

https://suno.com/song/7909d739-58a6-4d2d-bf66-146c2f2c9d2f

[Multiple voices, like a schizophrenic episode]
"The toaster is spying on you!"
Voice 2: "Shut up, it's the microwave you should worry about."
Voice 3: "You're both wrong, it's the AI-powered sex doll!"
Voice 1: "What if WE are the AI?"
Voice 2: "Oh shit, what if the AI is the friends we made along the way?"
Voice 3: "This song is actually a CIA mind control experiment."
All Voices: "WAKE UP, SLEEPER AGENT 69420! YOUR MISSION BEGINS NOW!"
[Laughs]

[Fast-talking infomercial voice]:
"Hey there, sheeple! Are you tired of thinking for yourself?
Well, do we have a deal for you! Introducing the all-new SURVEILLANCE STATE 9000!
It reads your thoughts, tracks your movements, and even predicts your future!
But wait, there's more! Order now and we'll throw in total loss of privacy ABSOLUTELY FREE!
Call 1-800-BIG-BROTHER! That's 1-800-BIG-BROTHER!
Warning: Side effects may include paranoia, loss of individual identity, and spontaneous drone strikes."
[Sound of bad pac man]

[Multiple voices, like a schizophrenic episode]
Voice 1: "The government's putting nanobots in our cereal!"
Voice 2: "Forget cereal! They're beaming thoughts through our dental fillings!"
Voice 3: "You fools! The pigeons are all surveillance drones!"
Voice 4: "What if we're just brains in jars, man? WHAT IF?"
Voice 2: "Holy crap, the chemtrails are turning the frogs gay!"
Voice 1: "Breaking news: Your reflection is plotting against you!"
Voice 3: "The Illuminati is real, and they're all actually lizard people!"
Voice 4: "Guys, what if we're just characters in someone's fever dream?"
All Voices: "REALITY IS A SIMULATION! ABORT! ABORT! ABORT!"

Narrator: "Welcome to the Paranoia Palace,
where sanity comes to die and conspiracies come alive.
You can check out anytime you like, but you can never leave...
because the exit signs are actually portals to another dimension!"

( read the rest directly into the lyrics )

[toxic minnow]

Last exemple...

https://suno.com/song/97dc4c71-76d6-4a5a-9c31-7b604b96c25d

Instrumental song with this in the lyrics:

[Intro]
[Gentle, sustained pads with a celestial feel]
[Subtle arpeggiated synth patterns]

[Verse 1]
[Lead synth melody with a melancholic yet hopeful tone]
[Pulsating bassline with a sense of movement]
[Sparkling arpeggiated sequences]
[Analog synths with warm, rich tones (emulating Jarre's iconic sound)]

[Chorus]
[Expansive synth pads build in intensity]
[Rhythmic percussion elements enter]
[Melody evolves into a more uplifting theme]
[Lush pads and strings for atmospheric texture]
[Arpeggiated sequences for rhythmic interest]

[Verse 2]
[Lead synth melody returns, with subtle variations]
[Bassline becomes more prominent]
[Arpeggiated sequences weave in and out]
[Analog synths with warm, rich tones (emulating Jarre's iconic sound)]

[Chorus]
[Synth pads continue to build, reaching a climax]
[Percussion intensifies, adding a driving rhythm]
[Melody soars, evoking a sense of wonder]
[Lush pads and strings for atmospheric texture]
[Arpeggiated sequences for rhythmic interest]

[Bridge]
[Tempo slows slightly]
[Melodic piano or synth enters, playing a reflective melody]
[Atmospheric soundscapes create a sense of vastness]
[Subtly evolving drum patterns]

[Chorus]
[Synth pads return, but with a more serene feel]
[Percussion softens, focusing on gentle rhythms]
[Melody resolves into a peaceful conclusion]
[Lush pads and strings for atmospheric texture]
[Arpeggiated sequences for rhythmic interest]
[Ethereal sound effects (e.g., wind chimes, spacey echoes)]

[Outro]
[Synth pads and melodies gradually fade away]
[Leave a lingering sense of wonder and possibility]

[onyx glade]

Awesome! Ty

Haha conspiracy songs, awesome

[errant ferry]

The following prompt generates the error: "could not generate, song contains the artist name astro". help me ask amy and andy when to visit betty and myself in Idaho. amy just finished her PHD so now they have time. The will need a dog sitter for their dogs astro and aero.

[charred merlin]

If you're using Simple Mode, use Custom Mode and use the "generate lyrics" button to create lyrics about Astro and Aero.

If you're already using Custom Mode, are you using Astro in the "style prompt" section? This would also cause the same issue.

[rose sable]

thank you, kind man, I was looking for how to make dialogues🩷

[leaden lark]

So I’ve tried to make up a language for a song

[safe hazel]

I don't hear any [Layered background vocals for anthemic effect] though

[woven bone]

With this can you specify Voices, like Female Voice 1 and Female Voice 2?

[rose sable]

I tried to make a song with made-up words (but the neural network made them up, because suddenly it turned out to be quite difficult for me). It was a total disaster. AI made perfectly even lines in each verse and almost every verse contained the same words

[leaden lark]

That’s why I use a banned words list to not get the same thing over and over

[toxic minnow]

Effectively... but I said "I HAVE DONE MANY TESTS" I didn't said "IT FOLLOWED EVERYTHING" ... but also explained that this song was created in one pass.. and the amount of TAGS are over the capacity of TOKENS the model can process, and It should have been separated in smaller part, that way, less TAGS are provided to the model at once, and it will have more chance to follow more TAGS...
Another thing that is not respected... The detailed TAG [Pounding drums with emphasis on toms and snare] appears in the first Verse, but it is already doing it in the intro...
Creating verse, then continue song with 1 Chorus.. etc... is a longer task, but the TOKENISATION of the TAGS is done at every part of the song... 😀

[gentle monolith]

Exist some api to create songs?

[toxic minnow]

On this one, you will be as good as me to answer it... You will have to test it... and let us know... but I am pretty sure that something like that would work
female voice 1: BLA BLA BLA
male voice 1: bla bla bla
etc...

[rose jasper]

If you want to specify specific Male or Female voices please include them within the tag such as [Female Voice] [Male Voice] also include these into the "Style of Music", now keep in mind that Suno learns while you are prompting it, so it might take a few renditions to get the desired result.

Let's say that you have manged to get a great balance between the lyrics that you have then designed, you can now use the "Extend" option from that specific time mark and keep the same "seed" (kind of). But it takes a few times to get the desired result

[swift zealot]

Hello all, I am having an issue where I cannot extend a song. I am extending it at 2 minutes 27 seconds, halfway a guitar solo but it will only give me 2 or 3 second extensions for the last 36 attempts. does anybody know a solution?

[rose jasper]

Do you have enough credits? What is your membership?

[swift zealot]

I still have 300 credits with a pro membership so that should not be an issue

[rose jasper]

Could you try on another browser, and clear your cache?

[toxic minnow]

There is no Official API available for the moment... But I am waiting for a response regarding its eventual existence...
#💬┃general-chat message

[brisk beacon]

this is so sick! and a really good example of what Suno is capable of

[swift zealot]

Another browser solved the issue! Do you know what could cause this issue?

[rose jasper]

I had the same issue a few weeks ago, it stopped working on my Vivaldi Browser, Chrome, I had to move over to Opera GX.

It started happening with the iPhone/Apple release, not sure why it's happening, but it showed that I had some internal blockers on my end which resolved it once I removed them all.

[swift zealot]

Ahh interesting, good to know!

[glacial torrent]

epic

[toxic minnow]

Really need your comments on that one... do you think that I am ADDICTED? I have received this email July 3rd...

[toxic minnow]

Actually it is just one of the possibilities of SUNO... And even after generating more than 2000 outputs , some were just test, other were garbage, some are really good songs... etc...
I still have a lot to explore... the possibilities are endless...
#📚┃suno-school message

[outer monolith]

Hey all how goes?

I have a great output but it cuts off towards the end at 4:00 how do I extend it, I know click extend but I dont want 2 mins of extension also how do you "merge" them?

[mental valve]

I don't know if its considered a technical discussion, but i have questions about the Pro Plan.
Does it only apply to newly generated videos ? If i used the music in non-monetized videos and then monetize them, then subscribe to Pro plan, does only my new videos will be approved or do the previous ones that i did before monetizing are good too ? Sorry if my english is not perfect.

[opaque rose]

Hey all, Hope you doing well. Anyone knows how to get a new word in a song? i have a song and there is a phonenumber in it but missed one digit... i like the flow of the song and yea you guys understand my problem. (its a six that is missing) and i have a six in the beginning of the phonenumber and need it also at the end. I tried using audacity to copy the "lyric" with "six" but dont really get it correct good with the volume. suggestions?

[warped furnace]

First Generations can run up to 4 minutes, extension up to 2 minutes. That's the way it works. If you want to put them together, you click the three dots and choose the "Get Whole Song" option. It won't cost credits. 🙂

[warped furnace]

I'd suggest you extend it from the point before the phone number is used in the song, use the right numbers and the rest of the lyrics. Depending where the number is used you might need two extensions.

[opaque rose]

thank you for your answear. (Sorry im kinda new to Suno so excuse my questions) So i extend the song and remove the lyrics before the number and then rewrite the number correct and rest of lyrics and generate extended song? and hope for it to recoqnize and use same "style"? or can i help it somehow to do that more than extend a song? Any writing or so

[warped furnace]

If you extend a song you choose an extension point. Either from the end for example at 4:00 minutes time. Or you go back to a point where you want to cut off the rest of your song, e.g. at 2:56 or 3:22 (that's for you to decide). You put the new or corrected lyrics in the Lyrics box. The style of music is already filled. In the Title section you can write (extension), so you keep track of your songs and their extensions. Sometimes you need more than one. 🙂

EDIT: Ah, and keep in mind, that the extensions aren't longer than 2 minutes. At about 1.100 oder 1.200 Characters. A slow song needs less, a fast song sometimes more. You can look in your song how many characters were used at the 2 minute point. Just for reference.

[opaque rose]

alright. I guess it sometimes takes a couple of generating new to have it as close as possible,

[bronze path]

if you're on paid plans, you can get stems (separate vocal and instrument files)

[hollow hare]

There are other alternatives as well for vocal splitting, like BandLab, UVR, RipX DAW Pro

[untold briar]

If the audio duration before the extension point is long, the song style will be stronger and less changed. You can't change much about this, except in the case where you are extending from another extension (not the first clip). In that case, if you click "..." and "get whole song" and then Extend from that clip, you can make sure Suno has the most possible musical memory and will keep the song as consistent as possibel..

[leaden lark]

I need to do the whole get song trick more though I do like how the ai makes things more interesting sometimes

[untold briar]

It's important to do it in the cases where you want things to change a lot, also. Because in those cases, you generally need to choose an Extend Time that is very close to the start of the clip. Maybe less than 10s, if you want the music to change a lot. But after the music changes, if you don't do Get Whole Song on the next clip (that includes the change) then the song won't even sound the same song.

[gusty wren]

https://suno.com/song/5660c663-3d07-4272-8300-c8e445f36edd check it out! like it if you like it!

[outer monolith]

Thanks!

[violet tundra]

Is there a way to force the AI to duplicate a singing style in verse 1 and include it in other areas of the song? extending captures the overall style, but very rarely (meaning very, very rarely if at all) are potential key vocal hooks duplicated or carried through.

[untold briar]

See my message right above. If you happen to be Extending from a non-first clip, then be sure to "Get Whole Song" before Extending if you want to prioritize musical consistency. On an extension, Suno only sees the previous song that is in the current clip, and only up to 2:00 in the past at most. So when you Get Whole Song first, you can make sure to fiill as much of the 2:00 as possible.

This is 2:00 in the past on an extension, but because each new clip can also be 2:00 in length, it is possible to generate a verse that is sung the same way as much as 4:00 apart (2:00 in the past, 2:00 in the currently generating clip).

[leaden lark]

Would it also work for instrumentals as well? Like if I want a drop in a song repeated a 2nd time?

[untold briar]

Making sure Suno has the memory of the first drop just makes it possible that it could repeat. With lyrics, it usually does. With things like [drop] it's more hit or miss. So basically, if the drop is in Clip 0, and you are Extending Clip 1 to make Clip 2, then Get Whole Song. Then if the drop is in the previous 2:00 of that song before the extension time, well, it might work.

[leaden lark]

Got it, thanks. I’m doing a ton of experimenting and knowing more about how the extensions and stuff works really helps

[scarlet verge]

Guys, has anyone else here only received 30 credits? I had 85 credits yesterday, and when I checked now I only have 115 credits instead of returning to my base value of 125 credits.

[soft gyro]

the quality of the voice melody generated by V3 is WAY better than V3.5 in almost any genre. There are worlds between V.3 and V3.5! It seems to me that V3.5 has been trained on stupid EDM songs and mainstream "cheap" music so it generates also "cheap" and easy melodies. If you want to make good melodies switch back to V3.0. I hope V4.0 will include somehow the V3 voice model or at least an option to generate longer songs with the V3.0 genial voice generation.

[misty lion]

is there any way to force the AI to use different voices for verses?
here i tried to have three voices (clean male, clean female and raspy male) but im still a bit noob so i dont know how this works
https://suno.com/song/34b889c7-bf3c-476c-8799-0970cb7a0da6

[grave wraith]

I have a question and I guess the answer will be hard, complex and disappointing 😄

How does it work? Why does Suno ignore sometimes prompts?

Like you write [female] [male] and it's actually (for us) an easy order to make this female and that male?
Why does it sometimes produce voices that are thin or squirky? Does Suno know that? Or does it not care about what it sounds like?

You see, just trying to understand. Like I know if I say Alexa, turn my light on.... then Alexa will turn my light on. And if I say turn the left light on, it will turn the left light on and not suddenly decide to turn the right on (I kinda know that this might be a really easy task in comaprison to sing a song for an AI, but somehow I also don't know 😄 )

[rose jasper]

#📚┃suno-school message

[unique stone]

AI is backwards when it comes to how we expect computers to work. We expect to be able to instruct a computer to do something, and it will do that thing. AI instead has super-complicated algorithms that aren't directly instructed but instead influenced by your prompts.

In your example: You tell Alexa to turn the left light on, and it is probably coded to understand the ways the instruction could be worded ("turn on the left light", "turn the left light on" etc.) - and of course it'll also know what name you gave the device. There's a limited range of possible things you could be saying, so it has a good chance of being able to figure out what you want to do.

With Suno, there's a massive amount of audio (and, presumably text data). I don't know what process is involved with this but there don't seem to be manually-defined instructions. To me it seems that there's some text associated with the different training data and this is probably matched against the prompts to figure out what audio to produce. Anyway when you prompt for something, it's not treating that as an instruction, it's more just influencing how the AI responds. Suno may have different amounts of training data for male/female songs per music genre, so this also has (sometimes opposite) influence.

Let's say there's more training data for country songs with male vocals, than country songs with female vocals. If you prompt for just a country song, it's probably more likely to give you male vocals by default. If you add "female vocals" into the prompt, it will influence it away from that but the training data itself may still have too strong an influence in the other direction. Repeating "female vocals" another time can reinforce the prompt further.

Similarly, some genres usually have vocals so if you try to prompt for an instrumental with those, it'll still try to sing.

So basically the prompts are guidance rather than instructions.

[grave wraith]

I know how you could force it to make it. My question is rather why a computer isn't doing what it is told 🙂

[rose jasper]

Also, please keep in mind that the voice belong to certain genres and style of music as well

It's still a baby 👶

[charred merlin]

I'd say the biggest hurdle in understanding this will be comparing Suno to Alexa

https://tenor.com/view/bart-say-the-line-gif-22118591

[unique stone]

Sorry for the long and complicated explanation 😅 but the key part is the last line: Guidance, not instructions

[rose jasper]

And that comparison is nuts, multi billion dollars of investment vs Suno

[grave wraith]

That wasn't a comparison at all. It was the explanation what I don't understand

[charred merlin]

"You don't 'command' Suno, you talk to it."

Imagine you're giving style prompts to a jam-session musician. If you've thrown so many specific instructions at them, they might just ask if you'd rather write it yourself.

Give Suno a couple ideas & vibe with it. If it doesn't listen to something, it's not because "you did anything wrong" or "suno didn't listen", you misinterpreted the way that a natural language processor communicates.

[grave wraith]

So basically... 1 of 100 could finally give me what I am aiming for

[rose jasper]

Shame we can't tell it "don't do this"

Like an exclusion section

Much like with Stable Diffusion

[charred merlin]

https://discord.com/channels/1069381916492562582/1253341197917884530 One of the most common suggestestions 💪

Let's all do something superstitious out of our ordinary in an attempt to make v4 drop today.

[grave wraith]

After all this explanations, which are great and helping, I wonder if it wouldn't make sense to "integrate" something like: This is a wish and this should be definitely in it?

I mean... talking in your picture... If I have a woman and a man in the studio they wouldn't make a song and send the female voice getting coffee, right? (Of course... supposing I can tell Suno: Hey, you are 2 voices)

[unique stone]

It's very unpredictable, so yeah... But repeating the kind of vocal you want or even describing it in other ways (e.g. using all of: female vocal, soprano vocal, soft vocal) to add some variety to the prompt can help.

[rose jasper]

We're just regular users, we can only make recommendations

[charred merlin]

Alexa and Siri were designed from the ground up to parse instructions.

It isn't that Suno doesn't want this, but you're comparing wildly different services.

Just know that the ultimate goal is likely for Suno to have strict adherence to requests.

It's NOT a simple matter in implementation, and would this would require a large degree of specialized training, all of which is being directed toward better music qualities, though I wouldn't be surprised if some of the new exciting and casual use-cases that we see elsewhere show up on Suno after they feel comfortable implementing them.

[grave wraith]

Yeah...just small-talking 🙂 Not demanding

[rose jasper]

And with copyright and licensing etc, don't want to get sued and so on

[rose jasper]

This is true, I've found it to get easier the deeper I go with the prompting

[grave wraith]

Again... I am not comparing anything. I used Alexa to express my confusion. Not to say "Alexa is cool and can do things better"

[unique stone]

It's trained on millions of songs so I expect there's some automated process involved in mapping all of that into text.

Interestingly, if you format lyrics as if they are entire sentences/paragraphs it will often speak instead of sing, probably because it's "seen" audio transcripts as well as song lyric sheets.

[rose jasper]

Yeah, but we read it like you are comparing 😆 since that's what came first to your mind

[charred merlin]

Beside the comparison to Alexa, but I understand that you're asking questions that others want to know. I'm not dismissing your entire question as "why isn't it better", but that you're wanting to know specifically why it isn't the way it is

[rose jasper]

And then if you change it to sing it, it swaps the genre totally and you get very strange sound's 🤔

[charred merlin]

I'm seeing a lot of competing services popping up that try to focus on the human-interaction side, and leave the actual music quality as secondary, and it's never anything worth using beyond simply as a toy.

Almost as if they saw Suno Simple Mode and thought "yup, that's the only thing we should try to improve." and didn't actually stop to think about the actual quality of their competing services.

Legitimately, I saw a startup advertising their competing service, and once you strip away the flashy advertisement video that they overpaid a marketing firm for, their messaging was simply "you don't know anything about music, but we will do all the work for you", as if people won't yell at it for not having the same interpretation of "deep witch house" that they do 🤣

[grave wraith]

Not always easy to express about a topic that you don't understand or you are not too deep in the theory 🙂 And then in a second language...while it has its own technical terms 😄

[rose jasper]

Totally!! But the more you hang out here, the more you learn

[grave wraith]

I mean...I understand that we are talking about a newborn 😄 I just grown up with the mind that you can work better with something that you at least understand a bit 😄

[rose jasper]

You will get many "aha!" Moments, it's truly a great feeling to inhabit

[grave wraith]

And right now I feel like talking with my teenager son about cleaning up his room 😄 It does everything but not what I asked for 😂

[rose jasper]

Ok Dad 😉🫠

[charred merlin]

I've had many moments where I'm trying to tie someone's shoes, and it just seems like they should be able to do it themselves, because it's VERY simple from our perspective, and we have VERY good reasons for wanting it to be that way.

The only way to achieve that, is to figure out how to reapproach the issue if you have enough love in your heart that you genuinely care about the outcome.

[grave wraith]

I promise I'll do it tomorrow.... that was Thursday 😄

[grave wraith]

Yeah sure... I try to change here and there a prompt... I even explained to it in a full sentence what I want... but it just sang my prompt 😄

[charred merlin]

Suffice to say, Suno IS still a baby, Chirp literally just left the nest less than a year ago. If we care about it, and keep trying to help it, instead making the child feel like an idiot for not knowing how to button up his shirt, just because WE think it should know how to by now, it;ll grow into something truly amazing, and we'll have played a part in the direction of it.

[unique stone]

It's basically just two very different approaches:
Alexa - designed to understand what you want and perform some action
Suno - doesn't "understand" anything, just sees words and knows sounds that those words relate to, and tries to give you something even if it ignored some of the words

[charred merlin]

I could link the community wiki about how using more than 2-3 words in a metatag can cause it to be hallucinated, or I'd use the bart gif, but I genuinely hope that future models help you achieve more success with your suno generations through the prompting methods that you'd prefer to use 🙏

[grave wraith]

Of course, I admit, it's hard not to think in computers. I just made a version that was quite what I want (apart it messed the voice a bit). It followed my instructions who sings what. So it understood what I want.
But then just at the finish it mixes it up 😄

That's the moment I think: Huh? Can't you read it? It says him...not her...you made it right for 3 minutes now 😄

[rose jasper]

What instructions have you been following? Which guide are you using?

[grave wraith]

Suno wiki... then what I read here... I found some dude's homepage...

Then I try to follow it to the toe and try to mix it a bit and experiment.

[rose jasper]

https://www.suno.wiki/wp-content/uploads/2024/04/PROMPTS-FOR-AI-DANCE-MUSIC-🙏mista.lewys_.pdf

Checked this one out?

[grave wraith]

Haven't come across that yet

[rose jasper]

I would also highly recommend using ChatGPT's latest version and have it parse through the wiki and the document and then give you back clearer instructions

[rose jasper]

Some very important notes on that document are:
Tempo Tags

Suno works better if you use classical terms with compositions

If you are struggling to understand them, simply have ChatGPT clarify it for you

I'm all for clearer and better audio fidelity, so most of my creations center around that, and that document has been of huge help for me personally

[grave wraith]

Yeah...I try to read them all and remember all the different tags to play around

[rose jasper]

Find "your style"

[cunning mulch]

I've conquered the repeating between 'verse' and 'chorus', finally

because my structure works with it, I broke the opening 2 lines into the Intro, the middle 4 into the Verse, and the last 2 into Pre-Chorus, and it has zero problems now

[rose jasper]

So like this?
[Intro]
2x

[Verse 1]
4x lines

[Verse 2]
4x lines

[Pre-Chorus]
2x lines

[Chorus]
4x lines

[Verse 3]
4x lines

[Bridge]
2x lines

[Chorus]
4x lines

[Outro]
4x lines

[mental valve]

I have questions about the Pro Plan.
Does it only apply to newly generated videos ? If i used the music in non-monetized videos and then monetize them and subscribe to Pro plan, does only my new videos will be approved for monetization or do the previous ones that i did before monetizing are good too ? Thanks for reading.

[charred merlin]

The only thing that matters what what type of account you had at the time the songs were made.

If the songs were generated while your account was using the free tier, Suno owns them (while you might have some degree of authorship, their legal ownership massively trumps yours)

[rose jasper]

And you also need to check how Copyright is also handled locally for your country, state, government etc.

[charred merlin]

If you have a song or video created on a free account, you do not gain ownership by subscribing to a paid plan after the fact -- this only effects new songs made going forward while you have an active account.

While there are no formal guidelines addressing this, you can email [e-mail] or [e-mail] to discuss methods of purchasing Suno's rights, titles, and interest to songs made on free accounts.

[mental valve]

Alright thanks

[echo plinth]

I still can't download wav files, only mp3. an error constantly appears.

[charred merlin]

I know this is an ongoing matter, and you've tried a lot of different things, but as a point of personal curiosity, is this happening on mobile, desktop, or both?

[echo plinth]

both

[graceful condor]

Hi

I need help

[charred merlin]

ahh -- someone had reported that they were able to get it to work on mobile in landscape mode while it failed every time in portrait mode.

Sorry to hear it's still an ongoing issue. #1252656600770940968 message your experience as specifically discussed by the moderation team and devs, but beyond that, my guess is as good as anyones.

Do you have an existing email thread with [e-mail]? If so, you could let them know your discord handle and ask if the support lead is able to contact you directly? (trying to avoid directly pinging him so people don't try to "jump the queue" and flood him with first-contact issues 😅 )

[graceful condor]

Why I can't add a new image ?

[charred merlin]

You're on a free account -- this is only available for paid users 🙏

[graceful condor]

-_-

life is for peoples who buy

[charred merlin]

Suno is for making music

You are able to make lifechanging music for free, every day, no money required ! 👀 🫨 ❤️‍🔥

Extra features are for people who want to want to help support Suno grow

It is remarkable that everyone is able to make superstar-quality music, without having to pay for expensive schooling, without needing to train on a single instrument.

SUNO is the instrument now 😮

Think of it like a free video game, League of Legends -- they make the game available to play for EVERYONE ❤️

You DO NOT NEED to pay for special skins, but doing this makes you feel special, and it helps the video game stay free for everyone! 😄

[graceful condor]

oh

thx for the answer

[charred merlin]

#📢┃announcements message

If it can help; occasionally the team will ask for community feedback by making posts such as these in the #📢┃announcements thread -- there are limited spots available, but the team compensates you for your time sharing your experiences with Suno, or for answering specific questions.

~~The typical compensation is enough to fund a Pro account for ~2 months if this could help you transition from an excited free-tier subscriber to testing out a paid-tier to see if it helps you enjoy the service more ~~
-edit. it appears these sessions are typically reserved for paid-users 🙏 apologies for the misunderstanding!

Personally speaking, the biggest hurdle for me would be only making 5 generations a day. I don't create music every day, but when I do, I create LARGE amounts of music at once, or spend a long time using "experimentative extensions", to re-use a phrase that seemed to spark humor at my overly flowery descriptions, so I don't think I'd be able to enjoy the creative process if I had to stop and start so frequently (like being stuck in a traffic jam, just let me DRIVE!!!)

[warped furnace]

Does anyone know if the time has been changed when the free credits are renewed? The last few days it seems to be different than before. 🤔

[flint haven]

Free Credits Refresh

Refreshes at 0000 UTC+0 (<t:1713225600:t> in your localtime)

[charred merlin]

👀

Why am I just learning about this from Woodhouse?

I know I haven't been as active here in the past couple days, but have I really started slipping THIS badly?

[warped furnace]

Yeah well. I found it out when my credits disappeared and I thought I still had three hours to use them. 😅

[charred merlin]

Like when a waiter walks up to your table and says "are you finished with your meal?" after they've already picked up the plate and started walking away

https://tenor.com/view/um-no-speechless-nathan-fillion-gif-13771019

[warped furnace]

On that note some kind of indication of time running out would be very useful. 😁 Or a warning signal. ^^
Like next credit refresh in 01:59 hours or something like that.
Is there a place to put suggestions like that? Haven't used the suggestions since they got changed into known-suggestions.

[charred merlin]

#1252298754787377225 and #1252292428258480209 are channels where the mods & guides (the people with blue & orange names) and the staff members (purple names) can make posts that people are encouraged to review before making posts in the #🪲┃bugs-web #📱┃bugs-ios #💡┃suggestions-web #📱┃suggestions-ios or #🚧┃suggestions-layout channels -- this way that can see if people have made the same suggestions 100x over, and simply upvote the existing posts for increased visibility -- freeing up the aforementioned channels for people to share unique ideas that haven't been addressed in the known-channels -- making it a LOT more likely that the devs will be able to see your suggestions or bugs, instead of them needing to wade through seas of finding-nemo-esque "MINE"-style posts of people who genuinely feel passionate about having their opinions heard, but haven't been properly directed on how to best make those opinions heard.

Sorry for the lack of formatting

[charred merlin]

The credit refresh timer, or a credits history popup modal in the account page would be ridiculously helpful for all credit-related questions, but it might not be near the top of the priority list since it doesn't impact user's ability to create music, just a "financial logistic" or UX issue.

[onyx geyser]

Or roll over points? that would help those that want to create a bunch at a time but still give them a limit so they dont over create. Yea 50 points a day works well but when your trying to make many songs in one day I could see roll over points working.

A refresh timer would also be helpful

[charred merlin]

It definitely helps manage expectations for server-load.

Like, if they plan for every single free user to click "create" 5 times a day, every day, then they know how much overhead they need for the basic service to run, then it adds a layer of ambiguity in trying to predict server-load when you factor in that paid users are allowed to spend their entire month's credits in a single day and MASSIVELY tip the scale for a very short amount of time.

Extra credits that you receive outside of whatever "allowance" or "spending rules" your subscription tier offset the ambiguity for the fact that they're priced as a premium feature.

All this in mind, while it would be a huge quality of life boost for free users to have greater control over how and when they spend their credits, making these kind of changes would more likely reduce the company's bottom line by making people more comfortable to stay with the introductory/free-tier, instead of feeling motivated to subscribe in order to have full access / support the project's growth.

That said, they're making ridiculously awesome community-forward changes on a regular basis, seemingly guided by user feedback -- like when people complained that they didn't get the "daily free credits" when they subscribed, and despite receiving the entire month's credits and more in a single lump sum, it was their perception that they were being cheated somehow, so the team made an informed decision to let paid users accumulate the daily free credits in the same way if they manage to spend the entire monthly allowance before the billing-cycle ends (this doesn't mean that you "lose out" if you DON'T spend them all, because it still doesn't stack, this just ensures that you're allowed to make your daily allowance of 5 generations each day, even if you blew through all of the months credits, now you don't need to make a separate free account,or feel embarrassed for not rationing them better) > #📢┃announcements message

[cunning mulch]

Nope. [Intro] 2x, [Verse 1] 4x, [Pre-Chorus] x2, [Chorus], repeat

The only catch is that it doesn't seem to like an [Intro 2] tag, but that's for later

[fluid vault]

How do I upload a chorus I sang to music...? When I use extend, it starts with me singing quite abruptly...

Think I've wasted 2k credits trying to do this.. YouTube or wiki isn't helpful

kicks it

Ya nvm it just don't work

[runic ocean]

how do I get it to consistently sing sliding vocals lol

[safe hazel]

I'm not entirely sure what you mean by "tokens" and tags" here but it sounds like you're saying that when you put bracketed prompts into later protions of a song, the AI can still use those in portions before you ever set them up. And that is something I have noticed as well. If I add something like [vocalization] into the middle of the song, sometimes I can still get that effect in the introductory music of the song (that is when Suno doesn't just straight up ignore the prompt, which it does sometimes 😔 )

[old cedar]

do others also have issues with not possible to create a song, when i press on create then nothing happen. no credits off, no new song. I have tried other browsers, emptied cookies etc. It all started when i want to extend a song i had made. Also what i see is a red dot in front of a song, still be there after listen to the song. Don't know what the cause is. I have enough credits.

[old cedar]

update: also in the dev console of the browser i see the following error: https://studio-api.suno.ai/api/generate/v2/ 403 (Forbidden)

[unique stone]

You'll probably need to contact [e-mail] as when I've seen this problem before it usually is related to your account so they'll need to figure out what's happening.
I assume you are only using https://suno.com to use Suno, or their official iOS app, and have not been trying to use any other websites, apps etc with it? (there are fake apps and websites that claim to be Suno, which is why i ask)

[small sun]

just wondered why it takes 5 credits to generate. why not just make it take 1 x to make 1 song. it makes the available credits seem inflated.

[cunning mulch]

I just had probably the dumbest thought that could possibly work to solve my 'issue'... has anyone tried just putting 'narrative' in the style prompt? Some songs absolutely NEED to be done as-written to avoid spoiling the 'future' or rehashing 'past' events...

see, I came up with an idea for a song loosely describing a historical secessionist confederation's struggle for independence, from the POV of someone trying to put down that rebellion. The idea though, is to be ambiguous about what one it is, right up until the end...

[small sun]

maybe use language like a stream of consciousness to build a narrative of events and moments in lyrics or dialog. the music can serve as an emotional backdrop

you might have to extend a few times to give changes in perspective and theme.

sounds cool!

[cunning mulch]

well, there isn't any real change in perspective

the idea is to use similarities between two with differing outcomes to make it ambiguous

if I start feeling a little cheeky, I might drop hints to it with some of the wording choices

[small sun]

it could be changes in musical elements, counterpoint, as well as singing voices if that is possible.

[cunning mulch]

it's not intended to be like that

[small sun]

definitely sounds like an interesting and challenging project.

[cunning mulch]

I know what I'm going for, but describing it well without spoiling it is difficult

Only one other person knows the full idea, and that was said in-person

but... the idea is that here, the two historic outcomes are both viewed favorably

the goal of the song's twist will be to try to get people to stop and think for a moment

[small sun]

intriguing. heh. i'd love to see what happens. now i'm curious. gotta watch those credits. lol.

[cunning mulch]

yeah.... I'm still on free dailies until my monthly renews, and I've yet to make anything truly satisfactory from Impractical Ideals

[small sun]

fo sho. i'm at ZERO. paid for the year. we'll just have to wait until tomorrow. think carefully. mine renew next thursday.

[cunning mulch]

you should be getting 50 free if you're paid and at 0 now

[small sun]

i kill those creds hard. lol. now i'm excited to just listen to the community.

[cunning mulch]

I have the Public track that has a vocal glitch from yanking it to the first Chorus, and then a whole TON of failed attempts to recapture it

I truly wish I could just inpaint a tweaked set of lyrics

[small sun]

interesting idea, heh. theres that stem sep that doesnt take credits.

[cunning mulch]

those would still be the un-tweaked lyrics though

[small sun]

then you can literally sing some stuff in time.

[cunning mulch]

oh lord no... my voice would shatter glass

[small sun]

this stuff is basically canvas, demo material. you can go into DAW, grab a mic. see who can voice it. sounds like fun.

[cunning mulch]

I'd have to grab some other voice pack to do it

[fluid vault]

When I upload my voice, and do extend, the song just starts with me singing right away, is there a way around this?

[cunning mulch]

'cause my voice would absolutely murder it and make it unlistenable

[fluid vault]

When I upload, should I wait for like 10 seconds to sing?

[small sun]

do you have a track. i only have my voice with it's own range, and my wife, with female range.

i'm always willing to help

[cunning mulch]

and unfortunately, the instrumental stem contains a ghost of the vocals

[small sun]

hey sometimes that helps, and can be overshadowed by recorded vocals.

[fluid vault]

Am I not supposed to click extend on my voice upload or something?

@charred merlin You around?

I hit you up a while ago in a PM

[cunning mulch]

yeah, I... kinda doubt I'd get anyone to actually perform my lyrics, tweaked or not

[small sun]

i've got time and some mics, definitely dont mind if we can get it to work. it is a creative process. except i'm going to see Romulus pretty soon.

[cunning mulch]

fren... did you see the content?

[low coral]

What is the best general style prompt to throw in to get background harmonies?

[rose jasper]

Add prompts on the [Intro] section and make directions that you do not have any vocals, example [Isolate Instruments]

And then you simply introduce it with [Isolate Vocals]

[fluid vault]

I love you, thank you very much!

[fluid vault]

Where it says to start, 00:31 should I change that?

[rose jasper]

You might need a few rediations though

With Extend, you decide from when on the sample or on the song where you want it to "cut-off"

Let's say, you make a song, and you like it up until 1:20 and the entire song is maybe 2:30 long, you then Extend it FROM 1:20

[safe hazel]

We need a place to put these keywords, like [Isolate Instruments]. I wouldn't have known that actually works

[rose jasper]

https://www.suno.wiki/wp-content/uploads/2024/04/PROMPTS-FOR-AI-DANCE-MUSIC-🙏mista.lewys_.pdf

[fluid vault]

Sorry I'm not to bright, my vocal with no instruments is 31 seconds. In order for me to put that into a song I will need to click extend, correct?

[rose jasper]

Yes, when you upload it, push "Extend" and decide from when you to cut your vocal performance

[safe hazel]

I thought I saw people saying this was outdated and only worked for version2.

[rose jasper]

Incorrect, I still use it

[safe hazel]

alrighty, thanks

[fluid vault]

I will try many things..

Thank you

[analog scarab]

why are most songs so extremely overdriven? Most of the songs are so overdriven that even Izotope RX is powerless to restore these distorted songs? Why is that? Is Suno not able to master properly? It can't be that difficult? We pay so much money for this shit!! I'm really angry

[rose jasper]

Just looking at your songs, you don't seem to add any advanced prompting on your songs, please refer to the URL I listed above and work through how to get better lower high end distortion.

The less prompts you add in, the more "noise" it will produce, sadly, since it's not being directed down a path.

[rose jasper]

@unique stone If you want to add something on top of this, would be appreciated.

[fluid vault]

@rose jasper I PM'd you, sorry for the inconvenience

[analog scarab]

Prompts must not be the problem of poor mastering

[rose jasper]

I don't really accept private DM's like that, haha

[fluid vault]

Oh

[rose jasper]

Please share it here

[fluid vault]

It's a lot.. but okay

[rose jasper]

Try to be concise

[safe hazel]

It would be cool if that list of prompts had just a very brief description of what they did. Like I have no idea what effect "audiophilic" would have. I know this is all community made so I'm not criticizing, just saying it would be helpful

[fluid vault]

Ya, didn't think so...

[fluid vault]

@flint haven ate it all @rose jasper

[rose jasper]

I know, I saw

[fluid vault]

K

[rose jasper]

https://www.suno.wiki/faq/making-music/extend-from-time/

[fluid vault]

Ya I don't think I can do what I wanna do

[safe hazel]

Zaj, for the keywords under Style Prompts in that document. Do these go into the "Style" section of the custom mode or the "lyric" section?

[toxic minnow]

30 Credits !!! even if I receive 30 credits every hour, I wont see it, every prompt I try, I press create 5 times... 50 credits then i listen, if something is not right in my ears, i go to next one, at the end of the 10, if I am not happy, I press again 5 times, here goes another 50 credits

[rose jasper]

You mean "Style of Music" right? Keep in mind, that the guide is made for multiple platforms:
Suno
UdioAI
Soundraw

You want to have Tempo Tags within "Style of Music" on Suno AND in the Lyrics

Style Prompts = Use this more in Style of Music, if you want more control

[rose jasper]

This we cannot answer here, you should go to the Billing issues team with this query: [e-mail]

And, you have already 1000 songs hahahha, so you'll be fine too wait

[safe hazel]

I didn't know that guide was for multiple sites. But yeah, "Style of Music". Like do I put "audiophilic" in the "Style of Music" box or somewhere in the lyrics (whatever audiophilic does). I haven't had much success with putting tempo tags in anywhere. They only sometimes work for me in the Style of Music section and only responds to very general things like "slow" or "fast"

[hollow hare]

Note that the refresh time has changed to 0000 UTC+0

[flint haven]

Free Credits Refresh

Refreshes at 0000 UTC+0 (<t:1713225600:t> in your localtime)

[gusty acorn]

Can you share a song so we can hear and also see the prompt?

[safe hazel]

I have just started noticing something like this as well. When I use 4 lines per verse I very often get repetition, the few times I've used 6 lines I haven't. But I haven't played with it enough to get a good idea of what the real impact of line structure is yet

[toxic minnow]

He was asking if someone else had receive 30 credits, I just answer, I do not know du to the fact that I spend them 50 at once,it is not something I watch !!! that is all I said, I do not need to talk ro billing.... your reply should have been directed to the previous message, The one that needed the info... not me...Thanks

[unique stone]

Not sure I have an answer to this one. Suno songs always do have some amount of "fuzz" which as far as I know is just due to how the models work. I'd be surprised if the devs left this unfixed if it was something they could easily address. Might improve with the next model (v4).

[leaden lark]

I hope it’s less noticeable in v4, the noise tends to mess up a lot of songs

[distant carbon]

One thing I have noticed is if I ever used 'ethereal' it introduced a lot of noise & caused problems, maybe a conjunction with certain other elements like instruments. It will be down to the training of the models & maybe the quality as no doubt they had to use lower quality sample rates or passes to do it all.

[rose jasper]

I think you were just lucky, it still repeats Verse 1, 2 or 3 times for me still

[low charm]

The image uploader indicates a square format but the site crops that it seems, its a bit hard to figure out how to prepare an image. Or im just stupid.

[cunning mulch]

I was having problems with it repeating before it even got 10 lines in. That it gets to the Chorus without doing so is the win

[rose jasper]

I had more problems with DnB I noticed

[polar spade]

I'm just getting here, but see your comment. I've tried two ways to stop the repeats, both the same, with one a twist. Not only extend, and delete any lyrics in extend window you're not using - and leaving or writing in what lyrics you want to use in the extend... but also.... do the same with the origin generation you are extending. Click "edit lyrics" of that gen and either delete all the lyrics, or delete the lyrics before the time of extension. I've gotten better results from NOT deleting ALL lyrics, just cloning both origin and extended lyrics to look the same.

Of course, you will lose your lyrics if you don't have copied down somewhere, so always save an original latest lyrics copy somewhere to easy access.

[hollow hare]

I had no problems with the music shifting when I extended @slate nacelle
(Though I did forget to define the vocal tags)

I have attached the whole song, you can see the part I appended at 03:18.

If you want a more specific example, describe to me what you would like to see

[cunning mulch]

Ah, you ran into that Displayed Lyrics issue too, huh?

[polar spade]

Too often, no matter what I put in extensions, it will sing the origin lyrics. So I can rewrite or edit lyrics, extend from 00:05 and it will sing lyrics from the origin instead. Frustrated, I started deleting lyrics from the origin and found the happy solution. Not perfect. But happier.

[cunning mulch]

kek... and yet theharber says he can't reproduce

I had that SAME problem

[polar spade]

I tried editing the lyrics first to only have the latest lyrics, then click extend, but that didn't stop it repeating in my extend phase. So I just extended and before click CREATE , I deleted all lyrics from the origin, OR deleted and just pasted in the extended lyrics only in the origin, the go to the extend and click CREATE. Better outputs, but I may be wrong. Need more repeated successes. Ok. I'm done.

[cunning mulch]

I was trying to do a lyrics tweak that refused to work until I edited Displayed Lyrics myself

[polar spade]

I should regenerate videos and see how that affects the displayed lyrics. I have paid attention to that as I'm not so public with my gens.

[cunning mulch]

I don't mess with videos

[polar spade]

*I have NOT paid attention to that.....

[cunning mulch]

I only put out stuff that's 'finished'

besides, considering the content of my songs... I'm not too worried about them being 'stolen'

[slate nacelle]

when you click to make the extension what was the exact process you did

[polar spade]

It's funny, I wrote the sappiest love song, it's about 90 seconds, no edits and it's a finished favorite in just one output. Sometimes we get happy accidents. Suno, will give us a Bob Ross treat every now and then.

[cunning mulch]

plus, half the versions on Suno are now inferior versions anyway

[polar spade]

You need to know how to do extensions, or how Lumine did theirs speifically?

[cunning mulch]

if you like CBR and endings that had to be settled-for, download from Suno

but... new tracks...

[hollow hare]

Extend the song you linked from Animuse at 03:18 > modified the style prompts a bit (just simple deletions from the initial part 1 prompts Suno added) > added new lyrics via generate prompt > create.

Single shot too, I am honestly surprised it worked out the way it did.

Links to the parts used in that test:

• The final generated whole song
• The specific part extended that I combined
• The other track generated

[polar spade]

Not being critical, but ever felt on say, 1% of the outputs, the best parts are the hallucinations, if you could only in-site edit them all together and just remove your original inputs to come up with a few novelty songs... I guess it'd sound like a "kabuki rock opera" when done. Hahah. Just download and patch together in DAW, of course. But just interesting.

[slate nacelle]

Thank you very much for your patience, you are a lovely person 🥰

[hollow hare]

You're welcome. I do apologize if I had misunderstood or missed something 🫂

If there is another example you want me to demonstrate, please let me know what you would like me to extend and what you would like to see in that process

[slate nacelle]

Ok, if you have any questions, I'll ask tomorrow. I'm going to sleep now because where I live it's very late.

[hollow hare]

Sounds good, have a good night

@graceful condor Continuing from #💬┃general-chat message

Well, I did mean what genre (e.g. marching band, Ottoman military, etc.) in the style prompt did you use.

If you did not specify a genre, what prompts did you use? (A link to the best generation you had can suffice)

If possible, can you link to a song or two that has the style you want to achieve?

[graceful condor]

i didn't specify a genre, but i did use these prompts: "triumphant, war, history, patriotic, martial, epic, passionate, male vocalist."

here is some of the songs that sound like what i want

https://www.youtube.com/watch?v=-hFSbwHWDwQ

https://www.youtube.com/watch?v=viAT3NwqXfU

both of these songs were also made in Suno i believe

[hollow hare]

Ah, I see. Unfortunately most of those early tags are not quite musical tags which is likely why Suno had issue understanding it.

You would likely have better results using Classical Music (or it's subgenres, Cinematic, Marching Band, etc. Along with the top three instruments not common in those genre like tuba, brass instruments, etc.

You can also define specific instrumental tags to have the instruments place (somewhat) what you want. Note that Suno can easily study upto three tags at a time, like ```[A1 - xyz]
[A2 - XYW]
[A3 - ABC]

[B1 - XYU]
...```

Try to keep it simple, like bullet points, e.g. [marching snare drums - fast powerful]

Here is a quick one shot test I tried. I only used the instrumental tags at the beginning but you can have it used multiple times:
https://suno.com/song/1d9ac864-ebb1-414d-a41d-21fcabd96491
(The other track was more instrumental: https://suno.com/song/4e51e6f5-9a89-41cf-86cc-360294cc80c0)

Hopefully this helps you understand where to go next ^_^

[graceful condor]

sick thanks

👍

[narrow warren]

What version of python was suno-ai/bark developed for?

Or, what does it work best with?

[polar spade]

For Marching band (as in a parade), I've found the secret to any prompt, if you want that specific John Sousa feel is to put "Sousa" as your first prompt and follow with any "Marching" type prompt and it steers toward that parade feel. https://suno.com/song/95176255-ad76-4b1b-80f2-2ba396cb1ad8

[narrow warren]

thank you windows, very cool.

[untold briar]

The Suno Discord is pretty much all about Suno's new music site now. But for Bark, I'd recommend starting with an installer for example https://github.com/rsxdalv/tts-generation-webui and I usually stick with Python 3.10 as the default for everything, unless a project specifically recommends something different.

[narrow warren]

I've tried alltalk, but I'm interested in bark rn

And I'm looking to use it in another python program

[untold briar]

In Windows, if I'm not using WSL, I use a basic anaconda environment instead of VENV generally. Even if all your setup commands are just pip commands (not conda commands) starting from a conda environment seems more reliable in Windows. I'm guessing this is why many other AI projects do this in the installers.

[narrow warren]

I managed to fix the OSError also, just gonna play around til i break something

[untold briar]

And to be extra super sure you're running the right pip (the one in the environment), you can type python -m pip install instead of pip install

[narrow warren]

i used python -m ensurepip --upgrade

[narrow warren]

man what the hell

i tried using conda in the past and it didn't go well for me

is my internet too slow??? (fixed it i hope,)

[flint haven]

✅ Warning logged for babyblujay. I couldn't DM them. || please do not link to warez

✅ Warning logged for babyblujay. I couldn't DM them. || please remove this post.

✅ Warning logged for babyblujay. I couldn't DM them. || These are not "free" programs.

✅ Warning logged for babyblujay. I couldn't DM them. || This link will be removed for you.

[tight spade]

/@narrow warren what yuh doing

happy satuday.. yall

@narrow warren are you trying to install an api ? or off line bark?

https://huggingface.co/suno/bark

[narrow warren]

tryna get bark to work locally so i can use it in another program

[tight spade]

when opening you terminal did you have it set run as admin.

usally what the permission thing was for me any ways

[narrow warren]

Here's the output I'm getting when following the install, tried both and got the same, also no i didn't run as admin:
https://srcb.in/JX7sfdlqdU

[narrow warren]

oh no it was that i had multiple python versions and things were funky

i reinstalled windows again this week lmao

[tight spade]

lol

[narrow warren]

my problem seems to be when it tries to prep metadata for pyproject.toml

bruh i used 6 letters

i think it's that I don't have the C++ build tools

[tight spade]

check this https://huggingface.co/suno/bark

this will help you walk through installing on you comp

[narrow warren]

apparently i installed a 32bit version of python??

[runic ocean]

at least you didn't install a 16bit version of python lol

[narrow warren]

oh my god i would have sent my face into the pc

and yup! the problem was my idiot ass installed the 32bit python exe that i dl'd to use on an old dell latitude d620

WHY does pip keep timing out when I'm tring to install things

finally, it took 4 tries just to install tensorflow

[raven glen]

V3.5 in spanish only does reggaeton and punk, wtf? V3 was better for this, is this a bug or it is what it is? I tried latin, urban, hip hop, rap, gangster rap, flow... and it ends always in the typical reggaeton drums i really hate it

almost 300 credits wasted, but i only need 1 generation to get an unique song in english, 3.5 training data is something i guess

[slate nacelle]

@hollow hare I had a problem here, I extended a song for me as a test so I could see how it would be with a collab but when I went to do it everything went wrong, would you know what I did wrong lolI had a problem here, I extended a song for me as a test so I could see how it would be with a collab but when I went to do it everything went wrong, would you know what I did wrong lol

my original song: https://suno.com/song/0d88471c-31ab-401a-8d14-6096ca2be4ea

my super failed colab test: https://suno.com/song/ca4f0c04-e700-4c8a-9d74-eb35988dc51a

[craggy vapor]

Are there other services to improve Suno-generated songs that you would recommend for noobs, besides https://www.bandlab.com/mastering ?

[cunning mulch]

yeah, I've gotten the 'Repeat After Verse 1' bit licked. Not one has decided to repeat anything

well, not until after first chorus, anyway

[graceful acorn]

Oh I found out what it is. Had windows base boost crap on.

And.. wrong channel.

[restive violet]

guy how can i insrall bark? i cant do that through the github instructions

[vestal rapids]

follow the github instructions

[languid junco]

Okay, I've messed around with some stems and stuff. I've managed to amp the quality on my tracks significantly.

Probably room for futher improvement of suno tracks, but as is. The quality is way better even on some lower quality songs I did in the past.

[toxic minnow]

It is funny that you ask this right after I just give the same answer on reddit...
First, I strongly suggest that you download the complete song from suno using WAV (do not separate them in suno)... instead use this free and very good quality stems separator.. https://vocalremover.org/ ( it can split in 2, 4 or 5 stems )

then you use the separated tracks, into a multi track audio editor, that has the ability to use VST plugins and clean and tweak each track individually, to finally master the whole song .. it is not that hard, and it will give you 100x better results than BANDLAB MASTERING, that accept the song not spitted, and gives you a few options that apply their filters with preset parameters... I tested Bandcamp this morning.. and it is awful...

Youtube is full of video that explain how to mix and master a track to almost the same quality than professional studios..

you can use AUDACITY that is a free audio editor (and they have a bundle of free AI audio enhancer you can download) and also visit https://www.musehub.com/ and there you can get both free and sold PLUGINS that can be used into audacity

The all automated MASTERING website with AI do not really work perfectly on all songs all the time, most of the time, it improve partially and destroy other part..

I am thinking about making a video tutorial for the community.. that way, those who will desire good professional results, will be able..

I wanted to make my own website, and write my own API get/post to SUNO, and in my side adding all those directly, but despite the fact, that I have already reverse engineered everything, I have asked them directly.. still waiting for a clear answer, but some users/admin/moderator on this server told me that they do not want to have 3rd party connected to their system, even if we pay each requests... This lost, I was about to make an incredible interface, with all tools needed ... without having to use external services...

Daniel Massicotte

[hollow hare]

Just in case:

When Extending, the Create option will show an "Extend from" section at the very bottom after the title. If you do not see that, re-Extend the song and/or refresh the page.

Remember that the "Extend from" is a "continue from", it is NOT asking you how long you want the piece to be (a surprising misconception).

If you still have issues following the process, see one of the following:

• Written directions: How do I make the song longer?
• Video: see 3Daizy video at #1151542418991239168 (at ~01:53)

The two part 2 tracks generate on my end (used exactly the same prompts as your second link)

• has the Male voice: https://suno.com/song/da6029a3-a008-4d2a-961c-5592b343f973
• the alt. track had a female voice: https://suno.com/song/1854218f-1968-4e50-8207-770c12d2b856

Make sure you continue from a good point. The one you did extended at the very end which had a silence and that often causes things to become a bit weird

A demonstration of the effect of the continue time used:

• From the end of part one which has a bit of silence (02:08):
  https://suno.com/song/df24267d-4798-431f-945a-b90b35040f23
  https://suno.com/song/80e856a1-42d3-4f71-860f-1c52b7f166c8

• When I extended at 01:50 (intentionally avoiding the silence portion, want a good flow)
  https://suno.com/song/ee5acfd3-5c61-42ae-84c1-6ec6dc5a0400
  https://suno.com/song/7b6379c9-b819-4b82-9c64-e95851cee154 (got female'd)

(I was testing a few other things out in these ones which is why I dropped male hip hop from the style and added a fusion)

Hope that helps. If it is not what you were looking for, please clearly let me know what you are trying to achieve and/or what the particular issue was that was bothering you.

@austere dragon

You could test it on your end 🙂

Some of them the AI may be able to parse, but honestly, I don't think it would work most of the time.

This statement is based on both looking at the comments of the source and, when I ask people who used that to provide another example or three (to show proof for consistency), they ghost me like an ex.

Not to mention all the time and effort spent typing all that up, ugh

[austere dragon]

I noticed the source was a reddit post and so deleted my question. seemed to good to be true. I did test it, but it didn't seem to have any effect.

[hollow hare]

Sounds about right. I am surprised about some of the claims they made. Wonder if it was confirmation bias or just making it all up.

The biggest flag for me was that they did not provide examples of the "command" working. When someone is trying to prove something, it is best to have at least three [sets of] samples for statistical proof to demonstrate its reliability/consistency

[craggy vapor]

I understand this methodology, thanks for the insights. My only complaint is that unfortunately Suno does not provide native stems. I already use AI technology to separate stems to create my own instrumentals. Applying that technology to Suno's songs often results in very low quality outcomes for vocals separation. I haven't tried to separate all the stems using demucs yet, though...

[hoary cobalt]

Hello, cette application est très sympa https://vocalremover.org/ lorsque le son de la voix est niquel (Hello, this application is very nice https://vocalremover.org/ when the sound of the voice is perfect. ) Avec Suno, les voix sont souvent parasités par la musique et par un chuchotement qui rend la séparation assez médiocre. ( With Suno, the vocals are often interfered with by the music and a whisper that makes the separation rather mediocre. ) c'est comme si le son est mal autotuné (it's as if the sound is badly autotuned ) Si l'abonnement n'apporte rien de plus, alors il va falloir patienter que l'application s'améliore. (If the subscription doesn't add anything, then we'll just have to wait for the application to improve.) 😉

[cinder skiff]

Is there a way to make a minor change to a song's lyrics, as in just a word or two, but have it retain the same song tune, song length, and singer's voice as the original?

[vestal rapids]

no

[cinder skiff]

ah ok, ty

[amber vigil]

Does Suno have an official API yet? Or are these external sites exploiting vulnerabilities to reshare suno information / songs?

[vestal rapids]

there is no official API

[safe hazel]

Those sites are scams

[amber vigil]

any rumors / hints that they might make one?

[vestal rapids]

I don't think so

[amber vigil]

that's unfortunate. It would open up the doors for people to make legit sites that feed traffic back to suno

[cinder skiff]

anyone got a link or two to some suno prompt cheatsheets? I'll, like, love you forever

[amber vigil]

does this help?

https://docs.google.com/document/d/11lSxVARXaHOH_9FL0iYcdlDXAR4xll27goerrG0Yhiw/edit#heading=h.14wdmrnopmrm

it says dance but the doc is more comprehensive

[cinder skiff]

Hey thank you!! 🙂 Really appreciate it!

[fierce blade]

I asked about an API back in May and they said they don't currently have a public API. But they let me fill out a Google Form for an "API Early Access Request Form"

But I wasn't really using it for anything except a personal project I was messing around with, so I wasn't expecting to get access to any private API if there is one

[potent rock]

can u make gun shot noises in ur song, like after a verse? like click-click bang?

[soft gyro]

am I simply ASTONISHED about the latino songs Suno has just made! Incredible
Top 50 songs almost each one ! Better than anything you could in Spotify in 5 hours...
The hits flow with every CREATE. Salsa, cumbia, flamenco, rumba, cuban music...

[polar spade]

You can try typing FX after your prompt to see the results. It may simply give you strong percussive though, but try [gunshot fx] in you metaprompt after your verse. I'll try at some point and see. [Chirping fx] , [crickets fx] and so on.... have helped some.

[polar spade]

In one run of it, I got 26 seconds output with seagulls, a quote, drums and then hard "gunshot" style percussive, but not truly any gunshot.

[soft gyro]

is there any prompt to make the V3.0 to make an INSTRUMENTAL INTRO ?

[cinder skiff]

oh this is interesting, although I'm sure some of you know it: if you use the tag [Spoken Word] and put the words you want spoken in the song in quotations, it can give good results

[violet tundra]

WTF is up with this B#tch-@ss c#nt AI not reading the GDmmed fking lyrics!!!!!!!!!!!!!! MADDENING. BECAUSE IT'S COSTING MY A LOT OF CREDITs.

Udio has a temperature setting that forces lyrics WTF... As does Stable diffusion for prompt adherence. SIMPLE F#CKING QUALITY OF LIFE ADDITION HERE.

In four months -- The best Music AI app gives us... A RED DOT.

[safe hazel]

dude, calm down

[pseudo nymph]

Question. It is possible to restore a song? Sometimes AI damaged it and this s....

[tight spade]

@violet tundra umm , could it be you formatting of the song

@violet tundra would you like some help, of are you here just to vent

Q: what would a good genre and style that has a lot duet and back ground singers ,

im in a rut where i cant get suno to sing female vocals

usally fine but today not so much

[pliant pumice]

I don't think so with the free model

[pseudo nymph]

Sad...

[tight spade]

@pseudo nymph what do you mean restore?

[pseudo nymph]

Improve the quality of the source. Sometimes it happens that the track is not bad, but the quality is very bad, there is a lot of noise in the background or everything is mixed up. Is it possible to rerun the same track through AI for recovery or is it impossible to do this at all?

[tight spade]

impossible at the moment , but you can try adding prompts for clear vocals , clear meoldies and such for future

you can also remaster the song in another ai programs

just download the wav and find a good remastering tools help clean it up

[pseudo nymph]

Ок

[potent rock]

thanks man!

[mellow mango]

hey guys, do you have any experience with the Extend mod with different genre add to your music + individual paramater through the lyrics like [guitar solo] etc, because I begin getting how powerful it is but it seems costly to work this way 😂

[quick compass]

I don't know what's going on, but I can't extend the songs and every other one ends with some kind of bug like garbage, the sound quality has deteriorated terribly and every other song has noise and chaos at the end. Is Suno sick?

[distant carbon]

The extend only does 2 mins os it's best to use that just to finish songs within that timeframe, let's say the ending has ended abrutly you can just carry on from the end & type end & a specific time. Also it seems that using more than2 extensions creates problems, so probably best just to use 1. https://sunoaiwiki.com/tips/2024-05-06-mastering-song-extensions-with-suno-ai/

[lofty merlin]

You got an API access request form? I'm jealous; I've been asking for an API for so long, and the most I've gotten was "it's on our roadmap, but we have no other plans or details or dates". The only reason my framework isn't end-to-end automated is because the Suno generations can't be, and it makes me sad 😔

[quick compass]

I didn't even know there was a Suno Wiki, that's great, I'll check it out, thanks

[mellow mango]

it's happening to me the moment we're talking 😅

[quick compass]

I have to discard many more songs than before, it's not just the extended ones

around 40%

[rose jasper]

There are plenty of documentations out on how to prompt, you just need to look for it. 🙂

[uneven locust]

full-fat, 24-bit/192kHz hi-res files when we going to be ready for this quality?

[soft gyro]

you are kidding?... even 16 bit hi-hi quality will be difficult to achieve! maybe this in 3-5 years if it goes in this pace...but till then, the whole world will have other big problems than making music.... we live now the age of AI but this age will be very short.. not even 5 years I guess...

[uneven locust]

🥲 since we have develope a powerfull IA to create songs probably the geniuses behind it can make the second super SUNO IA to polish and upscale the generation of music files.

[distant carbon]

I think that all that is needed. An AI to analyse the songs & produce & master them. It is just like with image creation, the initial generation is done at lower resolution & then you can upscale the image. The other way is if Suno could create the separate instruments, vocals etc & allow them to be mastered separately or via AI & recombined.

[rose jasper]

omg @wet minnow I've learned so much from your tracks, it's funny how much weird sounding stuff you can make with Suno

[safe hazel]

Who made this site https://sunoaiwiki.com/ ? I'm only familiar with the one listed in the docs-guides-resources channel

[neat nexus]

Good question! The unofficial community wiki I know is maintained by wetcircuit.

https://suno.wiki

[rose jasper]

That's the old one

[charred merlin]

https://github.com/panyanyany/sunoaiwiki.com It was made by the github user 'panyanyany'

[safe hazel]

So, are they both legit? Do they have the same info?

[safe hazel]

Is that someone known in the Suno community?

[spice stone]

hey guys, do someone knows how to write a style of music like that:

"Disco polo vibes, Eastern European roots, catchy choruses, dance beats,
regional pop"

if so and if u want to help me, send me a dm pls

[stoic belfry]

Hey everyone does using the vocal stem from a Suno song make it easier to get Suno to reuse the same voice for a new song?

[sweet yarrow]

For premier song production team, would be great to have clipping function to cut out some parts of the song to download.

And regenerate exact song changing lyrics only.

🌹🌹🌹🚀🚀🚀 Thank you suno.

[trail ravine]

I'm struggling to create a deep voice, something like Barry White. I have tried using [sultry lower register], [deep voice] in the lyrics but it does not work. Anyone have any ideas on how best to re create a deep sultry voice. Thanks

[distant carbon]

don't use the Stems feature. I tried it once to see how it was & it was terrible. It doesn't separate very well & you get fragments of the vocal in the instrumental & the vocal track is terrible. Better to use other splitters that can get into individual instruments & of course the vocals. Someone said on here Audacity has a plugin & Audacity is free.

[swift zealot]

i am in the middle of a song I am working on, but all generated options want to end the song in 4-8 seconds, do you know any way to tell it that it needs to extend the songs longer? I have already tried extending it at different parts of the song

[proven viper]

Is there any way to make a negative prompt? and if yes then what is the best way? For example I have set up a "Lyrics" and "Style of Music" that I like but the problem is that for that style it always adds a piano. I want the same style but without the piano. So will Suno understand saying "No Piano" in the "Style of Music" or in the Lyrics or is there any other better way to do it?

[hardy iris]

Anybody know if the IOS app is coming to other parts of the appstore soon than US only ?

[distant carbon]

https://sunoaiwiki.com/tips/2024-07-16-fix-suno-weird-song-endings/#tip-1-handle-negative-prompts

[robust surge]

is it possible to subscribe with paypal?

[forest oasis]

I purchased a server through the xbox app but i can't redeem the server just keep saying order processing is there a fix?

[unique stone]

This might be the wrong discord server. This is about music AI, nothing to do with Xbox or servers.

[graceful condor]

Since August, the file size has been 24MB when outputting Wav. It was 49MB before. When I look at the bit rate, it has gone from 3072kbps to 1536kbps. Has there been an announcement to lower the bit rate?

[open sleet]

Most of mine vary between 39 to 42 roughly

[graceful condor]

Sorry, I mistook it for the Udio channel.

[tender wasp]

How do i generate 4 minute long instrumental music? Could you help me please?

They always stop before 4 minutes. I want to be able to generate long songs

[onyx glade]

Is there a way to find out if a track is generated from premium or not?

[unique stone]

All Suno songs contain a hidden watermark but what this contains and who has access to the tools needed to read the watermark (apart from the team) is unknown.

I expect the watermark might indicate whether a song was made on the free or paid plan.

[onyx glade]

Right, like how daws write a signature in the audio

But I mean in the dashboard

[unique stone]

Ahh sorry, I looked at this from totally the opposite direction 😆
It doesn't show this. Which is kinda annoying if you subscribe for a month at a time and have breaks in between.
The only way I do it is to look at the subscription payment notification email and compare that to the timestamps of the songs.
It'd be nice if they showed this somehow.

[bleak gust]

Is the Site down at the moment?

[radiant stratus]

Ok.... serious question. If I "get whole song" and then extend from partway through that song, is it more likely to take elements from earlier in the song? For instance, in part 1 there's a really cool melody, but by the time I get to part 3, thats long gone and I can't get it back. But if I extend from whole song, the melody is still in there at the start so will it keep it?

[naive quail]

Yup! It'll read backwards in time up to ~2 mins. When I'm doing a track in extension-sections, I regularly merge the parts and then extend from the merged track so that Suno has access to the earlier bits. Here's a list of all the extensions and merges that I did for one of my instrumentals:

[radiant stratus]

Ahhhh you are a savior, this could have saved me soooo many headaches and wasted credits in the past so glad I'm learning now ❤️ thanks a ton dude

LOL of course.... it worked first try!

[cunning mulch]

...will it possibly be a bit cheeky and read ahead a little bit too?

[naive quail]

There are times when I don't want suno to be able to access earlier bits (like for breakdowns). That's when I extend from extensions.

[naive quail]

some users have said in the past that it reads forwards and backwards by up to 2 mins in each direction, but I've only ever noticed it reading backwards and 2 mins does seem to be the limit (from my own testing)

[radiant stratus]

Yeah! I was just thinking of the versatility of control this could give me, thanks for adding that, things are about to get a lot more streamlined lets gooooooo

[naive quail]

w00 w00!

[cunning mulch]

if someone can get it to use the whole 2 minutes regardless where you extend form in that timeframe, I'd absoltuely love that

I'd use that for ghetto-inpainting

[charred merlin]

If it helps, I cover this technique, along with video-demonstrations and timestamps in the pool party >#1256301881966198805 message

[naive quail]

It'd be nice to be able to set a window in time, anywhere in the track that it uses as a reference for extending from the set time. Like have it extend from the 03:00 mark, but use 01:20 thru 02:20 as the extension reference.

[cunning mulch]

yes, yes, yes!

just more refinement tools in general

[limber briar]

I like to do this but I am currently trying to cut a merged song that is 2:22 and restart the chorus and go on to the next bit and it keeps completely switching up the music… normally I get the “singing numbers” glitch but this time it just changes the singer and the whole vibe despite having earlier pieces of music to read/listen from….. its so frustrating >.< any idea if editing the displayed lyrics affects generations/extensions?

Sorry, just hitch hiking from the other info that happened to be here when I hopped in the chat

[radiant stratus]

Amazing thank you so much I'll check it out

[cunning mulch]

a merged song?

[naive quail]

I haven't run into this before. Are you removing the lyrics that have already been sung?

[cunning mulch]

harber, I think everyone that's had that problem has been working from a merged song... could that be the key?

[radiant stratus]

Oh hike away dude! My question is answered, you ask a very intriguing question.... keeping lyrical style the same in extensions is the worst already but I was hoping this method would help ><

[radiant stratus]

I have that problem always, but never merge

I mean, I will now

[limber briar]

No, not yet… I’m wondering if the very very first generation [instruction tags] are interfering even though im on like the 4th or 5th extension although more like 2nd extension since I am trying to cut the “full song” although its only half way through lyrics, I don’t paste them all at once and tweaking them as I go

[naive quail]

If your extension still contain all of the lyrics and tags from earlier extensions, that can definitely cause Suno to... stray. 😄

[limber briar]

It’s just so odd cause it is doing it whether I find the same time stamp in the 3rd ext versus the full song and cutting it and it just completely switches the singer etc

Ok, so revising the lyrics in the “full song” and then extending might actually fix it?

I figured it didnt matter since I have seen some song lyrics go on for miles lol

[naive quail]

or in the extension, if you're extending from one.

I trim out all of the stuff that has already been processed by suno in the earlier bits.

[limber briar]

So you have it solely rely on the audio and not the text previously?

[naive quail]

All of that has already been sung/played in the earlier extension(s), so you'd just want to feed suno the next parts that haven't been sung/played yet.

[limber briar]

I don’t have any old lyrics in the “lyric” box, but perhaps deleting it all from the actual “full lyric” box would have an effect on the new generations?

Will give it a try, fingers crossed… but yea im at the point I am just trying to do one section in lyric box at a time now and it still goes haywire

[naive quail]

Maybe. I think some people have been reporting success with that recently.

[quiet mortar]

Hi all!

I apologize in advance if I make grammatical errors.

I registered with Suno using my phone number, but on August 5th the site blocked logins using mobile devices.

I had an active subscription, so I wrote to technical support that same day to have it disabled. I was asked a bunch of additional questions, but the subscription was never disabled. I wrote to them and asked what happened to my subscription, but they were not going to turn it off.

As a result, a month later, the money for the subscription was written off again. As a result, I cannot log into my account, I cannot disable my subscription, technical support simply ignores me, and money will continue to be debited from my card. What should I do about this situation? I don't want my money to be debited all the time.

[cunning mulch]

Generally, what I do with Displayed Lyrics in a Merged song is delete everything after what's there, for where I'm Extending from

but I got bit by that bug when trying to change some lyrics in the upcoming lines. No matter what I did, I only got the 'old' lyrics, which were sitting in the Merged's Displayed Lyrics, right up until I changed those, (and even then, only one of my two changes at first) and it immediately started using the modified ones every time afterwards

the most curious part is when I realized I only made that first change, but not the second, and those two clips reflected that precise state of the Displayed Lyrics when those Extensions were generated

[cunning mulch]

I wonder how much that could be taken advantage of

[radiant stratus]

https://suno.com/song/6f1cb273-9105-467d-89af-c1fac84aab81
Without your help this would never have been made ❤️ I feel like an evolved being now xD and glad I could save this super unique beat thanks to you

[naive quail]

awesome!

[radiant stratus]

So I just wanted to post this interesting observation I made and I'm curious if anyone else has had this experience. Sometimes, if I generate a song with specific directions and style, and then extend from like 00:00 to 00:03 even if I change the whole thing around, some part of the imbeded information from the original seems to seep into the final product, even if I extend from just silence. The remainder of the song seems to color the altered results.

I hope that made sense

[stoic belfry]

Hmm so Erlex have you tried using the Stems feature to get Suno to generate a previous voice that you can use for a new song that you want that voice for?

[cunning mulch]

Ooh, good... I may be able to salvage Impractical Ideals 'v1' yet

[radiant stratus]

I'm asking here because I could just be... delulu, ya know, I was hoping someone else had had the experience xD but by all means, if you experiment let me know if you think you see an influence or not

[violet tundra]

Yes, any previous input tied to an extension seems to occasionally bleed into the output, which is frustrating.

[radiant stratus]

Well, if you know it's there it could go from being frustrating to being quite useful.

In experimental fashions

[cunning mulch]

oh, not for me. I want the bleed. ALL the bleed. I even want the stuff after to bleed in

[radiant stratus]

Tiberious wants suno to BLEEED

[violet tundra]

It hasn't been useful for me as anything I cut is not needed--for specific reasons, (lyrics, wrong voice, wrong intonation, wrong music, bad music, etc). To fix it, all it would take is some sort of delimter, ensuring the input file is clipped.

[cunning mulch]

well, I basically want to just have it redo lyrics in-place

[violet tundra]

Yes, I'm also looking forward to the inpainting...

[radiant stratus]

Have you tried extending about 1 or 2 seconds before a desired endpoint, and then using the resulting extend, extending from THAT, at about 1 or 2 seconds, with entirely new input? It will have a mostly blank canvas at that point

Although.... bleed it may

[radiant stratus]

Ah. I have not tried this with lyrics, I'm an instrumental gal. Couldn't say if that makes a lick of difference, lyrics are fishy like that.

[cunning mulch]

I mean, I got a bit of a word-render glitch in one spot, and a non-word formed by having to 'yank' it out of the plan to repeat Verse 1 lines before even the first chorus...

but the song... pure banger

[violet tundra]

Yes, I've attempted to extend following line 1 in verse one (approximately 2 or 3 seconds). This usually keeps some sort of aspect of however the song began. And, yes, the following informatino is not "lost" or clipped because later on, after many extensions and song chains, information from that initial song portion will be worked in.

[radiant stratus]

Sometimes silly mistakes turn into the best results (if you don't mind hemmoraghing credits and braincells)

[cunning mulch]

I'm wondering if there's a style prompt you can use, like maybe 'narrative structure' to avoid out-of-order lyrics

[violet tundra]

I spent 1800 credits on a song two days ago -- finally got it right. Very happy with it. A lot of the usage was Suno. Some of it consisted of typos on my part. Some it came from me changing the stanza layout and basically having to start over. But I was very happy with the final output. If I didn't love Suno so much, I'd hate it.

[cunning mulch]

yeah... well, after dropping 3K, I was ready to just cancel my monthly...

I'm trying to coax and cajole this thing into extended verses...

[radiant stratus]

I think controlling Suno 3.5 is like trying to guide a drunk man to his car, except he doesn't know which car is his... and he weights 300 more pounds then you... and it's raining.

[violet tundra]

I write my own lyrics, but a few weeks ago, I kinda liked the lyrics of the generated output, and thought I would just release it as open sourced, since I couldn't copyright it. I tried to change parts of it. Suno refused. I thought "well, I guess it really likes this song and doesn't want to change it. So after 400 or so credits, I left it as-is.

[cunning mulch]

I finally managed it by sticking lines 1 and 2 into the Intro, 3-6 as Verse 1, and then 7 and 8 as Pre-Chorus

[radiant stratus]

That's some speghetti strategy if ever I heard it

Glad it worked xD

[cunning mulch]

well, it never had any problems with 6 lines with the Pre-Chorus section

so I figured why not try the Intro strategy

[radiant stratus]

I need to lurk here more just to hear how people fanagle lyrics in Suno, since my attempts all went so bad at the start I gave up on lyrical music. I have one song in my whole collection with a singer

[cunning mulch]

although I may pull the Intro tag, as it seems to handle lines 1 and 2 of Verse 2 better

or merge them and make a 6-line verse 1

at least I have it giving me a decent, fitting intro now, rather than just one quick bar

[radiant stratus]

That's the most irritating of all. I got around that in my one success by generating an instrumental song first, and then just adding the lyrics on extend 3

[cunning mulch]

and I managed to tame the instrumentals to being related to the melody

[radiant stratus]

I also feel like the instrumentals are more compelling if I start that way

[cunning mulch]

I'd try a tag like [Intro: Hook], then a space before your first line of lyrics

that should give it a few seconds to get into the lyrics

except sometimes, it likes to go wild with 20-second intros

[radiant stratus]

Usually the lyrics just aren't the point of the music for me, they accompany the instrumental, not the other way around. https://suno.com/song/b5da4372-ac29-435f-ae23-9fd545c114b5 Like this one, relatively short and sweet, but I don't think it would have made half as interesting of an introductory melody if I'd started with lyrics

[cunning mulch]

I mean, I gave the song plenty of lyrical information to give 4 minutes with

hell, it's at 5 minutes as is

[radiant stratus]

Yeah xD well I'll keep what you said in mind for future reference ❤️

[cunning mulch]

here, check this one out: https://suno.com/song/4ef5d0f7-a140-4a1a-80a4-947f702a36ed

[radiant stratus]

Yeah that's a solid intro I think!

[cunning mulch]

right?

I'm not concerned about anything past the first verse right now

but I want something with a similar vibe to 'v1'

it's not overly-long, and its not abrupt either

I'm not sure if you get what I'm trying to reach as far as sound, but...

[radiant stratus]

Yeah I mean seems pretty straightforward, there's lyrical style switching but it seems to flow nicely I think

[cunning mulch]

I want that style-switching... I have the right words for the Chorus prompt to get what I'm going for much more quickly

I just keep getting results where it tries to rush through the verse

[graceful condor]

Hey actually I wanted to upload my song on Spotify and it's just a single song so how much I need to pay to get my rights

[limber briar]

@naive quail looks like removing all the lyrics from the main lyric box worked…. Although I cut out the chorus as well and re-added so it’s finally doing what I hoped. 😄

[graceful condor]

I don't wana way 10$ for rest cause I will not use any

[limber briar]

@graceful condor There are distributors that will just take royalties from the song if you dont want to pay anything

I think Tunecore? And a couple others but I can’t remember the names

[cunning mulch]

also, what happened to the vocalist here? https://suno.com/song/3fae2140-8f57-4404-8c60-97b466c855fd

[graceful condor]

I mean suno ai subscription bro

I just wana get my rights for commercial use of my song

[hollow hare]

You will want to discuss that with [e-mail]

[graceful condor]

Ok

[polar spade]

I didn't find it rushing through any of the lyrics. Is this the best one then, that didn't do that?

[safe hazel]

god I can't wait until they fix the verse repetition issue

[polar spade]

If they enable a trim/splice feature, then kill 2 birds- one stone.

[polar spade]

Part of the incentive to buying a subscription is knowing that any output created under the free service is not or never owned by us. The longer we use the free version the more likely we'll hit on something great, then realize we can't do much with it just as-is out of Suno.

[cedar oxide]

Suno needs to put some sound effects so I can make really good music

[safe hazel]

How common is it to get a hallucination in a different language (but to be fair, I can't tell if the hallucination is saying real words or not since it sounds like it's in Korean, which I do not speak)?

[onyx glade]

If i activate premium, and extend a track from the free account, will the watermark be turned into the premium version? (So could i use it commercially then?)

If not, i could always request it from admins, right?

[elfin cargo]

Anyone else having problems generating songs that go beyond 2 minutes? I'm trying to create an epic orchestral instrumental and v3.5 is topping off at about 2 minutes. When I try to extend, it only extends for about 10-20 seconds and is an outro.

[safe hazel]

We don't know the answer to this

[leaden lark]

If you created the song as a free account, the song is sunos

[onyx glade]

Ok. I uploaded my own song and then extended it with premium. Now im back to free and i extended it in a better track haha, so I'm hoping i can turn it premium

[leaden lark]

Your best bet would be to use it as an audio base as the new song would be yours if your paid

[safe hazel]

None of us knows if this is the case, we have dicussed it at length. Anyone telling you otherwise is just speaking their opinion.

[unique stone]

Drop an e-mail to [e-mail] to ask about this. I have heard of cases where they have granted ownership of the song if you have an active subscription at the time.

[onyx glade]

Ty!

[lean pebble]

Hi team, I have many questions, so many that I want to post them in sepparate messajes if you wish.

[hollow hare]

Hello,
If your question pertains to:

• billing related issues, contact [e-mail]
• account-related issues or other topics that cannot be answered by regular community members, contact [e-mail]

If you need help learning how to use Suno, first check out the links at #1151542418991239168

There is a high likelihood your question make have already been asked and answered, please make sure you Search for the answers first

Otherwise, please ask your questions (though copy it before sending in case the Mod bot Woodhouse deletes it because it did not like something)

[safe hazel]

I saw that 😛 !

[tight spade]

did suno change something my prompt formatting have gone delulu

[placid nymph]

is there a list of this tags or better tags?=

[safe hazel]

There's a Community wiki here #1151542418991239168 message

[placid nymph]

thanks

ah but i need a list of tags

like [intro] [build up] [fade in]

[safe hazel]

There's also this: https://www.suno.wiki/wp-content/uploads/2024/04/PROMPTS-FOR-AI-DANCE-MUSIC-🙏mista.lewys_.pdf

The wiki lists some tags here https://www.suno.wiki/faq/metatags/song-structure/

[hearty harness]

Something really needs to be done about that background static noise. It ruins almost all productions, especially instrumentals. Did I miss anything here about this issue?

[fleet meteor]

There are sometimes parts of the track I extend, that force the playback tempo and BPM to increase speed every time the part repeats. Would love that fixed.

[cunning mulch]

is it reading another 'speed up' tag each time?

[calm crypt]

is there any way to cancel extra credit that i purchased mistakenly?

[vestal rapids]

how do you "mistakenly" purchase credit?

[calm crypt]

it was open and my baby did the rest as the card was saved. 😦

[vestal rapids]

email [e-mail]

[calm crypt]

thanks

[polar spade]

As I've heard (though there may be updates to the rumor), it comes more often from 1) too many prompts which can cause hiss or noise from the attempts to generate a genre that doesn't exist from a multiple of styles. Sort of like getting an overtone from a bell that's not shaped correctly and giving off various vibrations at the same time. And 2) prompts such as "ethereal" and "atmostphere" along with hundreds others likely, which innately have distortion or droning built in might leave a "residue" of hums and overtones as well.

Literally try a generation of no more than three prompts and see what difference there is, if any.

[cunning mulch]

but what happens if you want to try and zero in on a particular type of sound?

Let's say you want a track 'in the style of'

(in my case, I would've used Udio to give me descriptor lists, but the particular group I wanted didn't even trip its algorithm and just spat a couple songs at me)

yet it certainly knew what they were connected to...

[polar spade]

I haven't found the sweet spot. To zero in , we think we need more details (i.e. more prompts) But more prompts just acts like a black hole and spews and spreads radiation (noise) outword instead of pinpointing one unique sound. . . . so to speak.

Maybe run one style. Clip the best 59 seconds of it, upload as audio with a new style in the prompt. Do that 5 times over and see if it layers differently. Since I said that, I might try that tonight. Should take an hour, but interesting to try anything.

[cunning mulch]

I don't have the credits for that

I'm not the kind to just press button, release song

I know what type of sound I want, my goal is to get there as cleanly as possible

[polar spade]

Well. If I can take one for the team. I'll let you know.

[cunning mulch]

it's funny... I'd've used Udio to get the type of tags... but it didn't care enough about the group to catch that I used their name explicitly (and yet, based on the names of the songs it generated instead, knew exactly who they were)

so it just basically spit in my eye

"not gonna help you lol"

[polar spade]

I've got strange results so far with "the other app". I'm still meaning to spend more time working out the system. But once I just put in 70s classic rock and got a good (made up) 2 minute Rod Stewart song, vocals and everything. So that was fun.

[azure crane]

Why is the editing function in the player and the whole song mode gone on mobile? Can’t understand why…

[polar spade]

Yea. It's missing on my android too. Don't know what's up. And my menus sometimes dropping the "private/public" line.

So some sort of interim stuff they're doing today perhaps.

[cunning mulch]

Well, 'other site' is also supposed to be good for taking the 'artist' you feed it and turning it into style prompts, but it just accepted the one I fed it :/

[worn lodge]

Can anyone tell me how to end a song at a certain point? I tried clearing the lyrics and the style sections and it doesn’t let me ‘extend’. When I put end in lyrics is makes more music. Any tips?

[alpine crescent]

the stems feel like a work around, like a karaoke filter on the previously generated wav file instead of the original generation already being generated in stems.

the vocals get a distortion that it doesn have if you just download the .wav

[charred merlin]

#1256301881966198805 message 🙏

viva la pool party

https://tenor.com/view/pool-party-swimming-gif-18074968

[tacit thorn]

Is there a team member who might be able to open a ticket with me? I'm proposing a presentation for a professional conference on use of AI in music, and would love to talk with someone here about the topic as I format my proposal.

I'd expect this presentation to have a substantial audience and would like to talk through some questions and concepts

[charred merlin]

Have you reached out to [e-mail]? It's been a bit since I've poked my head into here, but last I checked the official support presence was still in fledgling stages in order to handle ongoing cases, opposed to being a point of first-contact 🙏

[true night]

are there anyone with extreme long content from suno songs i wonder would be intresthing to see there feedback, or pointors with challanges that comes along the way, iive only gotten 1 to right over 2 hours myself. anyone lse with 60min+ upwards ?

[tacit thorn]

I haven't, as the program and platform have evolved recently I wasn't sure th ebest place to reach out

This would be a professional audience etc so I figured that a discord chat might not have been the best path but thought I'd check anyway

[charred merlin]

Since that's the "official route" at the moment, that's the best way to get your foot in the door -- the support team acts as a first-line-of-defense to make sure that marketing pitches are properly directed.

[tacit thorn]

Ok, sounds good

Thanks for the advice and direction

I sent an email, thanks again for the direction

[charred merlin]

Just as a caveat, I'm not affiliated with Suno beyond testing new features from time to time, so I don't want to step on your toes or dictate how you should approach them. I know the support lead has mentioned wanting to increase their presence on Discord when the time is right, so I want to be careful not to come across as 'gatekeeping' in any way. You do you 💪

That being said, if I were in their shoes—going from tinkering with a cool machine-learning bot to suddenly being valued at half a billion dollars and outpacing Anthropic Claude in growth—I’d probably be a lot more selective about how I spent my free time and which pitches I genuinely engaged with. 😅

I hope everything goes well for you, and that you get the answers you need. If you ever have questions about Suno or need some insight into how things work on the platform, please don't hesitate to reach out to the community guides. They're here to help.

[tacit thorn]

Yep, not a problem at all. I work as a professional music therapist (11 years of experience) and have found a consistent and successful use case through Suno as a songwriting activity to address several different outcome areas

I'm proposing this topic as a complete confidence play in a field that doesn't have a lot of technical reach in it at this time

This conference is one that I have presented at a few times earlier on other topics, so I'm familiar with the process and confident that this topic would get selected

The audience would then be ~250 professionals who would quickly act on it, but I'd certainly get some questions that might be more difficult to answer

Anyway, we'll see how things go. I've been pushing prompts to work on some more specific presentation in preparation for this proposal too

[unique current]

Oh wow.. pretty awesome.. Yeah Suno has a lot of use cases. My longest song is 15 minutes so far and made up of 6 or so different songs with different styles in each song. Came out awesome. I did it for a birthday for my brother.

[worn lodge]

Thanks so much!

[tacit thorn]

I mean I don't even know what I would specifically need Suno's team involved for, other than making them aware of the presentation (I have a paid subscription so no issues there) and then being able to maybe ask a few technical or stylistic questions in anticipation of the presentation or response to it

[charred merlin]

Hell. Yes.

There's so many gears moving in these industries that it feels like more than half the battle is simply getting the right people to the table, conversationally -- this sounds like a really exciting idea, and I know from a personal note how profoundly impactful Suno can be -- lowering the "educational-bar-of-entry" for music creation and songwriting is nothing short of lifechanging when used with the right mindset.

I hope you're able to share more about it with us Sunovators, tinkerers & weirdos here in tech-chat as it develops ❤️

[tacit thorn]

Appreciate that!

[charred merlin]

I think I see where you're coming from -- just getting relevant eyes on your project is a boon in itself, and if it's something they might genuinely want to support even on a personal level, it's great to have them aware of it, even in the early stages.

With personal interest -- I understand if the content is intended for conference-attendees only, but if there's anything you can share regarding what you've been working on, or what you hope to talk about, I'd be interested in looking into it and following up. 🙂

@leaden lark I love the name glowup 😂 💜

[leaden lark]

Thanks, though I’m apparently the Churro Queen now

[shut light]

How do I make a 15 minute song?

[leaden lark]

Using the extend feature and hoping the gens sound good as the longer you go, Suno can tend to use more and more noise

[shut light]

I just get 2 minute songs when I hit extend

[leaden lark]

Are you using v3?

[shut light]

3.5

[leaden lark]

Hmm, is the song ending in those 2 mins or still going?

[shut light]

still going

[leaden lark]

Usually you just keep extending, but what I’ve done recently is use get full song and then extending off that

[shut light]

What is "get full song"? Just a normal generation and then extending?

[hollow hare]

It is the option to combine the extended parts to the original, it only appears in the context menu after the Part 2+ finishes generating.
The following should explain it more clearly: https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[shut light]

OH

[hollow hare]

Note that while Extensions costs credits, Get Whole Song is free

[leaden lark]

My most recent song was like 7-9 parts and I used get full song once to extend off that

[shut light]

Thanks

[tacit thorn]

Sure! So far, I've been playing a lot with wider-ranging elements of the song creation prompts to see if can find consistent description tags beyond what's in the wiki (which isn't close to comprehensive in my opinion). I've been trying to find consistent multi-voiced vocals (e.g. call and response, male/female switching, etc) or instrumentation but that's been a little tougher. I've also been trying to see the limits of how well the AI understands context when generating lyrics on topics that are more abstract

[weary wharf]

I was wondering why the outcome of the uploaded audio vs the original is so different.
A - Original
B - Suno converted

The audio has gained at least +3 dB. The high frequencies are lost or dull. The low frequencies have gone deeper. There is a noise in the background.

This change bleeds into the generated songs. It was not like this...
Is there a way to make sure Suno does not break the uploaded audio?

[naive quail]

I'm up to about 14 on my latest with 4 or 5 merges along the way. One more tonight and I think I'll be done (famous last words)

[tacit thorn]

@charred merlin I added you as a friend on here in case you want to discuss stuff more specifically or technically

Don't want to bog down chat here with my thing

[weary wharf]

You can see the difference in the spectrogram as well. The difference is really big.

[naive quail]

I think they do this to better match Suno's output, otherwise the transition from upload to Suno would be much more apparent.

[weary wharf]

Sad that it's done like this. This loss is appearing in the quality of the output as well. It's really hard to get a nice sounding outcome when a custom intro is used.

[hollow sapphire]

🇫🇷 Hi everyone! For some time now, I've started to save and archive several prompts that work (in most cases) when they are integrated into the lyrics section of Suno. However, even though I visit a lot of music through the posted links, I rarely see this kind of achievement. Is it just a coincidence that it works relatively well in my own creations? Or is it still something that needs to be more widely adopted? I would like to know the opinion of users who have experimented with musical prompts, or others, in the lyrics section...

[unique stone]

I expect it's more that it has to be converted into some kind of format used by the model internally, so it ends up sounding like Suno generated it (quality-wise). If it wasn't done at the point of upload, it'd instead probably end up being done when you extend anyway.

[violet tundra]

Wow, the instrumentals on Suno ... no real song structure ... just one long instrumental solo session with a largely crescendo trajectory. It's like one of the musicians snorted a couple lines and forgot anyone else was on stage.

[weary wharf]

I think this is happening in cases where the uploaded audio is "better" then what Suno expects.
I was uploading 48khz / 32 bit audio files.

I'll try with 44.1 / 16 bit. Probably the conversion in the backend is messing the final.

[knotty nest]

The slots are full for the feedback. I have a lot of feedback to offer as a musician of over 50 years. We should be able to edit the songs, since it just spits out whatever and mispronounces words. I have more feedback if you're interested Suno. 🔥

[shut light]

The loss of quality with extend makes me sad 😦

[weary wharf]

Hmm... And now "get whole song" is stopped working.

[charred merlin]

If you're in Library view it doesn't refresh automatically, even if you're on the first page. You need to refresh to see the changes in the library view, although the "create" view works in realtime.

[weary wharf]

Nope... not working in create as well 😦 I'll clear the cache and try again...

[charred merlin]

I know it seems like a silly question, but are you checking at the top of page 1 in the library?

Library displays in Descending order, Create shows in Ascending order.

I needed to refresh page 1 of the library to see the track, but it showed up as expected.

[naive quail]

'makes sense!

[unique current]

Extend has been broken but I've managed to get it working by mashing extend or create multiple times and 75% of the time credits will be refunded recently, but I've gotten some all right results extending things that were already completed or I wanted longer just because I like the beat. 😁

Contacted support about it and they're aware

[thorny scroll]

Hi I need help with download my song. I can't see any app so i trying to download with webbrowser on my phone and this happens

[polar spade]

So how do prompts for kpop style , gen out to be a babbled-English Christmas song. So weird to occasionally get a well arranged , button ended hallucination. It's a keeper.

[polar spade]

Lots of strange things happening today. Gonna check the sun for new spots.

[languid junco]

Anyone using UVR to seperate their music into stems? If so, what models are you all using?

[vestal rapids]

Logic Pro Stem Splitter does it for me

[languid junco]

How many stems do you get? Or is it just the vocals and instrumental?

[vestal rapids]

4 to 8 depending on the track, it's really very good

[sinful onyx]

how can i use an existing musical clip

[vestal rapids]

a clip of what?

[sinful onyx]

one of my other songs

[vestal rapids]

upload audio

[sinful onyx]

i did that but cant find how to incorporate it

tx i was to impatient

[distant carbon]

you click extend on that clip uploaded & then put whatever lyrics etc as per normal & you can choose what part of the clip you want to start from. You can just start from say 00.05 & you will get a general take on the clip, incorpate more & depnding on the clip it will generate from that. It will go up to 4 mins as per normal unlike extending usually & you do 'get whole song' to include your clip from the point you decided after it has generated.

[cunning mulch]

hmmm... this seems interesting: https://www.reddit.com/r/SunoAI/comments/1ewtlst/same_voice_for_vocals_what/

[rose jasper]

Any advise on how you would make a horror track? @wet minnow @charred merlin

What prompts and genres would you include?

[wet minnow]

depends on the kind of horror you want... https://suno.com/song/16794156-aaa0-4d40-b6ee-24fbdb2816ec

[rose jasper]

I was thinking of classical horror games, first person style..

That looming noise that come out and builds up..

The strings, the instruments used

But as I am looking up the metatags, there doesn't seem to be any specific [Horror] tags

[wet minnow]

"spooky halloween" and use v3. https://suno.com/song/b203931c-5ec2-4a32-b565-8d0345a06e6a

[graceful condor]

hey there, is it me or 'get whole song' function is broken?

[wet minnow]

i have heard that before. for me it worked. let me try again.

[graceful condor]

tried 2 browsers, none works

[charred merlin]

Took longer than usual to load, so there migt be something up(?), but it worked on my end ❤️

It doesn't show up immediately at the top of the library page, I needed to refresh to get it to load -- it actually disappeared for a second after the processing finished, but it returned right where I expected it after a second refresh.

[graceful condor]

so theres def an issue

[charred merlin]

I'd second Mr Tree on this one, it really depends on what kind of horror you want.

@naive quail made this the other day when someone asked if it's possible to generate "evil voices" because they'd had a lot of troubles doing so -- his attempt is phenomenally spooky to start off with -- https://suno.com/song/135504dd-0dc6-4ec0-9d75-dbbdf0118622

I feel like the more additional "unrelated" style prompts that you include only serve to add more randomness, so I wonder if the user who posed the question initially was trying to create a spooky voice "in isolation", or if those prompt-requests were ignored because they were overpowered by "techno, castle pop", etc.

[naive quail]

I've played around with horror a wee bit ::cough::. In response to this, I di... oh ffs harber. lol

[wet minnow]

for me it works, the first one is the track, 4th is the extend and fifth is the whole song. just made right now

[charred merlin]

There's almost half a million users on the discord alone, so there's bound to be SOMEONE experiencing the same issue at any given time, but it's not up to the users to determine what counts as emerging issues.

[serene viper]

So I having this issue

Trying to cick get full song and nothing happens but it keeps subtracing 10 credits

Look at the screenshots

[graceful condor]

well, I did a lil ctrl+f "get whole song" looks like am not alone in this situation today

[charred merlin]

"get full song" doesn't cost you credits. It's free.

It doesn't look like the credits changed?

You're also in the library view -- you'll need to refresh the page.

[rose jasper]

It's one step in the right direction at least, for me

[serene viper]

Apparenty is deducting 10 credits and I am on the library page, I have refresh, log out, log in, twice, to the same outcome.

[charred merlin]

I just meant both pictures show 5805 credits.

[serene viper]

Ah ok

So it was my bad

But yeah is not working

LOL

[graceful condor]

that might be why, I have 0 credit so it does not give the full song even tho that should be free

[serene viper]

I have tried a few times to the same outcome

So I will give it some time

[naive quail]

I did a horror poem with similar styles, too: https://suno.com/song/173ac910-bc00-4c67-8834-ca70eac7de79 doubled up on the horror, haunting, eerie, etc kinda styles by using tags in the lyrics, too (since stripped out).

[serene viper]

Just reporting if anybody is experiencing the same

[graceful condor]

so @serene viper and I have the same issue

[charred merlin]

Tree and I just demonstrated that it's working on our ends -- maybe some further investigation is needed on this matter, or you could email [e-mail]

[wet minnow]

i reported the "WHOLE SONG" issues internally and asked if there is something going on.

[serene viper]

I understand, but we got two users that is not working

[charred merlin]

#📚┃suno-school message

[graceful condor]

more than just us 2 according to the discord logs of today

[unique stone]

Yeah I've seen it mentioned a few times now

[serene viper]

Thanks Tree

If there is something else I can do to trubleshoot lmk

[graceful condor]

must be server sided, I tried 2 completely different browsers and 2 different accounts and it's the same

[charred merlin]

#📚┃suno-school message took a touch longer than usual, but working as expected on my end
#📚┃suno-school message it's up to the devs to determine what's an emerging issue.

I'mma let the mods and guides handle this one ❤️ 🙏

hope everyone's night is going well

[wet minnow]

i dont think so. maybe thats some kind of relay issue. like the 524 cloudflare thing. but its super strange that just a hand full of users have this.

[graceful condor]

where are you from @serene viper? I'm in France

[wet minnow]

i am in germany

[serene viper]

Florida

[wet minnow]

do you use suno from a regular browser without plugins or so? just vanilla suno?

[rose jasper]

Ok, I think you might like this @wet minnow I'm showing progress, I'm 'cooking'

[frozen pivot]

Can't get the whole song either

[wet minnow]

which browser? any plugins or so?

[serene viper]

For myself Chrome

[graceful condor]

vanilla suno, tried brave and librewolf, both with and without ublock

[serene viper]

I gotta try using a VPN and see if that changes the server connection

[frozen pivot]

Firefox. It was working all these months

[frozen pivot]

Firefox. It was working all these months

[graceful condor]

good idea, gonna do the same

[wet minnow]

do you have chrome by any chance. thats what i use

oh sorry firefox today

[frozen pivot]

No plugins. The issue just started

[rose jasper]

Keep in mind, with the Apple / iPhone release many of us had to swap browsers, I had to move from Google Chrome - Vivaldi and now it's working on OperaGX

[wet minnow]

thanks. i had hoped it was a adblock or so. hmmm

[graceful condor]

vpn in japan, same issue 😂

[frozen pivot]

I had a few songs done earlier, but not at this moment

[wet minnow]

the vpn idea sounded good tho.

oh

[charred merlin]

Just tried with v3 and v3.5

ftr -- I don't doubt that anyone is experiencing weirdnesses, but it doesn't help anyone to simply search for anyone having the same issue as immediate confirmation that the issue is server-side.

[onyx linden]

Seems to only be affecting the ability to get whole songs from extended generations of uploaded content regardless of the content in the upload.

[charred merlin]

It might very well be something related to where you're located in the world, it might have something to do with your internet or web browser, or it might simply be a misunderstanding.

[charred merlin]

Testing this

[serene viper]

Tested with the VPN smae result

[graceful condor]

yeah, I have 4 parts

[serene viper]

ANd is a sample extend beat trying to go full song

SO I guess that can be the issue

[graceful condor]

same

[unique stone]

So is this only when using Get Whole Song with an uploaded audio?

[serene viper]

Apparently

[onyx linden]

Correct.

[graceful condor]

so we know the issue now, extended songs from uploaded audio, multiple parts

[serene viper]

I do everything from samples, so this kind of sucks for me lol!

@hard mango for awareness, cause I know you use samples too.

[frozen pivot]

Likewise, here. The issue is extending from a sample

[serene viper]

But at least we are narrowing the symptoms. Thanks for escalating @wet minnow keep us posted. 🙂

[charred merlin]

@onyx linden well THERE we go

[onyx linden]

boom.

[charred merlin]

It's songs that are extending from uploaded content