#📚┃suno-school
1 messages · Page 61 of 1
your lyrics greatly influence the output
less so than in previous versions
but it is there
can the AI chop?
100% agree with this and the following post -- it's almost a Suno mini-game in my mind to hear something WILDLY unexpected from my prompting, then trying to critically assess WHY it made that. I had a silly rap song that took on a blues vibe, and then I realized it was because I said "woogie woogie woogie" as backing vocals.
9/10 songs turned out reasonably how I expected. One of them misinterpreted my request as "boogie woogie".
I bet you know which track I ended up keeping.
I'm not quite sure if I understand what you mean by chop -- do you mean a specific style of sound, or do you mean to end the song as a particular point, in a DAW-type interaction to remove a specific drum sample (like we chatted about in gen-chat)
rapping, chopping. Eminem in Godzilla and Rap God for example
when he raps really really fast
like TechN9ne too
I haven't tested it myself, but I read that combining-your-words-with-hyphens-makes-suno-treat-the-spacing-differently.
Makes sense in theory, because separating words to new lines acts as a method for spacing them out, so trying to "remove" spacing by stitching them together like this could work.
Actually, I'll test it out right now & share
example: turning "pop rock" into "pop-rock" helps it distinguish it as one genre so i've noticed.
if it's not hyphenated, you end up with a pop song, a rock song, or a pop-rock song. Like a box of chocolates, you never know what you're gonna get.
At least I've heard in some of the members songs rappid rapping. I have NOT heard any , hmmmm, what would you would assume as chopping. But keep in mind, we humans chop in order to impressively pronounce every word with a time frame. It's difficult and takes practice. With suno, there's no problem. So as suggested, write the words phonetically, with or without spaces, with or without hyphenated, and suno will simply pronounce it all.
Some members, as I have, have written phonetically in order to create an accent, along with prompting. As for rapping, between prompting for rappid , easycore, hyper and so on to style the song, write the lyrics as "chopped" possible to see if that improves the speed.
Lorem Ipsum isn't nearly as funny as a rap as I would have thought.
I feel like I just stole the Pope's airpods
what about how to prevent the lyric from repeating itself?
so, suno tends to switch to Italian phonetics when it encounters Latin...
It often loops some lyrics even though I dont specify it
thats just a generic AI problem.
Unfortunately I don't really think there's much to be done about it, but sometimes it ends up working out quite nicely.
And of course search in suno for "chopped" songs. Here is one just prompted "rap" but got a faily fast speed as many words are typed in one line, which pushes suno to fit all the words into a meter. If you search "chopped" most than not I'm getting "chopped and screwed" which is a different style altogether. So keep in mind, the word "chop" now has about 7 subgenres.
uk drill, rap, song. Listen and make your own with Suno.
Sorry thers the rap I found.
Literally nothing. All you can do is assume Fox Viewing Position, and repeatedly bang your head against the Generate and Extend wall
damn
be prepared to blow through 4K-5K credits if Sunos' going to be specially osmium-skulled
and as for lyric free longer moments in a song?
Bet to use something like {instrumental] tags
4-5k credits....but there's not even unlimited 😛
I prefer something like [Instrumental Hook] or [Extended Hook], so the solo doesn't go on for days
A curse? A bonus? No one knows. But many I've used to make new songs. It's getting pushed to have some trimming tools in the future, I think.
trimming, inpainting... basically anything to work with songs we already have
I notice that it tends to skip or repeat lyrics more frequently the closer you approach the character limit -- as if you're overloading how much it can think about at once.
As other noted, until we have inpainting as a feature, you need to go back before it sang the wrong lyrics, create an extension point and continue from that spot again to try to get it right 🙂
As a general rule of thumb, the more specific of a song you have envisioned in your mind's ear, the harder it'll be to get it as a "one-shot generation" through the use of complex prompting.
I have a workflow guide where I share a lot of my experimentations > #1256301881966198805 -- I've been focusing elsewhere since it got bot-heavy, but there's still some helpful tips in there 😄 Also, if you want to discuss ways to collaborate on getting your desired results, I'm open to helping out -- we both get to learn from the experience 😄
speaking of 'overloading'
yes?
I'm having an issue, not sure of someone here can help. I have almost 3000 credits but when I click on create its telling me i need at least 10 credits
heh, the music didn't master properly so it just hides all my lyrics under the noise
For instrumental sections, you can turn on the "instrumental" toggle to try and prevent lyrics from being used 😄 You can combine the different extensions afterward -- this way you have granular control over when the lyrics start and end -- instead of trying to control it through complex prompting, like "[extended guitar intro], or "[vocals start at 30 seconds]", which might be more likely to be interpreted as lyrics since it overloads how much Suno can interpret as single metatag as.
so I can decide when extension starts and it will erase everything else that was there before?
it erases after that point
after what point?
the time point you Extend from
#1270928841614823546 message - "how to extend" tutorial video.
Keep it mind it doesn't destructively edit, you only stitch things together in a new track. All shown in the video linked above.
#💬┃general-chat message also described extensions to another user here
so I can select it
yes
you can cut off words doing that though, as the AI will not shorten anything before the point to fit your new clip
now, if you want something Suno MIGHT recognize as instruction... if you want a line to start at a particular place, I tend to go with [Vocals Begin Beat X]
let's see again
Hey! I probably can help with this one!
https://discord.com/channels/1069381916492562582/1263857907011813517
The single known-bug I was able to confirm when I was on the guide team 🙂
If this ends up being the case, please upvote the post in the link so the devs know it's still an issue
what if the "rigid memory" issue on Extensions is caused by it being 'overloaded' by trying to Extend from a 'Full Song', rather than a 'Part 1'?
I'd like to help, but I'm not sure I follow -- can you explain what you mean by "rigid memory issue"?
that thing you couldn't find any way to reproduce
Yes. I think you got the general idea. As tiberious says.... erases anything after the time-set point you choose. Choose your point in the song you want to extend or splice onto, note that time on paper , click EXtend in the songs menu, then click create to generally find the songs lyrics up , scrolling down to seel the extend "timeclock" where you then type in the splice time where you want to now extend from and "record over" the following lyrics to that time point. It helps to remove any lyrics you don't want, in the lyric box, and of course, paste in lyrics that you DO want which should (but not always) start after that new extended point of time.
okay thank you i tried logging out then back in, i will change/update my payment method next see if that works
Hello
Or just follow that video....
Thanks your an excellent to generate music platform that Suno
Suno is a fun app to generate music. Yes.
@polar spade #📚┃suno-school message I'm seeing that you and Tiberious discussed his issue with lyrics entered in the "edit lyrics" section appearing in the extensions as hallucinated lyrics.
I spent a genuine amount of time trying to help him reproduce the issue, and it always came down to "well, it works sometimes and doesn't others", so I had NOTHING to assist with or to even try to use as a foothold to explore the issue -- I can't help if I'm unable to reproduce the issue.
I'm not trying to be dismissive and insinutate that there's nothing to be discussed, but I genuinely have no idea how to reapproach the matter.
If you're able to shed light on any testing methods that can cause the words entered in "edit lyrics" to be spat into the ensuing extensions (without being included in the extension-lyrics, specifically only edited lyrics in the origin section) -- I sincerely accept any advice.
I can tell Tiberious cares about this issue, and I genuinely have no idea how to help him, as any attempt to find a common ground to base experimentation from has failed.
the "kekekeke, harber says it never happens" just demotivates me from putting anymore heartfelt effort into it, but I don't want to leave a genuine inquiry on the table.
İ haber a few queetions , İ would be happy if you Could help me
how it can sometimes get completely locked-in on going by what's in the Displayed Lyrics field, and wholly ignores what you may have altered in the Lyrics Input field when making an Extension
you came back and told me you couldn't find any method to make it happen all the time
let's see now
all set thank you
Can we load Any instrument sound into Suno and produce a track on it ? İf yes, how?
And yes, it is important to me... that bug damn near made me cancel my subscription over it
if it wasn't for the daily-50 lifeline, I probably would be back on Basic by now
I recognize that you're very close to the issue, it might be that complicates trying to approach the matter objectively.
Every bit of "evidence" you've shown has been a 4-comment reddit thread, or using someone else's anecdote.
It's not that I don't WANT to help, dude.
I tried to explain carefully how best I work, and if we're not able to collaborate productively, it doesn't mean that I don't care about your issue -- it's just not productive for us to argue over how we think the tests should be conducted.
for technical feedback on specific sections, highly encouraged ❤️
https://suno.com/song/1d8db652-d878-4f63-9d92-50506f8e35e6 well this was one of the two results when it finally became more...understandable. And then this one was the other one https://suno.com/song/c9c4a4f1-04b9-476e-b4a8-2b4664c43d39
The second one I feel was better. It's a rather short one
I've tried different combinations and sequences with the "edit lyrics" of the origin generation. I can't get anything too consistent to say it's helpful. The extended sections always have something bewitching to them. When it works as wanted, I think I've done something right, but then can't repeat the success. So suno is just pulling my chain. 😊
If I iextend it I think I can have it keep building tempo until it reaches 160 bpm
It's going for cyberpunk/industrial/trance feel
Are you asking if you can upload any external instrumental? If so, yes you can (as long as it isn't copyrighted music)
When that issue, plus v3.5's utter stone-headedness about its idea of 'structure', combine to drain my account of not just half my 2,500 credit allotment, but also all of another 2,000 credit top-up, it is going to make the issue close to me
Might not be quite as imagined, but you can upload a 6-60s recording as an audio upload and Suno can continue from it -- it's not quite "magic" or the type of flexibility of audio-training tools like Elevenlabs etc, but still very powerful to start a song 😄
@safe hazel thanks , how ?
do you see this bar?
Does 'Sound Art' in the genre tags help encourage it to use your upload sound. I've heard it's meant to.
In DM, when I said I didn't know how to assist you further with the matter, you asked me what you should do when you experience the issue again.
I would genuinely suggest contacting support@suno.com if the free community-provided assistance is failing you to such a degree, and I apologize I wasn't able to be of more immediate assistance 🙏
(there is a 60 second maxiumum to uploads though)
I've had some really cool songs that implemented the uploaded sound as part of the beat.
Can't say I'm familiar with this -- is the suggestion that "sound art" in the style prompt section causes it to "focus" on the audio upload section, instead of leaning more on the improvisational side of things?
I genuinely hadn't heard of this, but I'd be interested to tuck it into my back pocket to take a look at 🙂
Same tbh, and even made some
of my own sound even better
so any feedback on https://suno.com/song/1d8db652-d878-4f63-9d92-50506f8e35e6 ? I mean I spent 10 minutes on the lyrics cause I put a timer on it just for the heck of it
Futurepop dark synthwave, catchy choruses, Cyberpunk Layers, Subtle Industrial Fusion song. Listen and make your own with Suno.
but regarding the prompting etc?
it's only got about 1 minutes 10 seconds of actual song
@safe hazel When we want to EXTEND after the music iş created ,the resulting alternative 7-8 seconds, why? When you say EXTEND , shouldn't it take longer to re-do the Song? Or is this a credit/ paid job EXTEND future?
I haven't noticed any issue with just uploading audio and letting suno do it's thing to need to nudge it more. Are you saying, for instance, if you simply upload an audio , let's say 59 seconds worth, and ONLY put "sound art" in prompts and , I guess, either one of ustom or instrumental toggled with no lyrics, it would spit back out the same audio with maybe just some fidelity enhancements?
Here's one of my favourite examples of Audio Upload being implemented in the beat. https://youtu.be/1IjDGHiW2qA
Genre: Oh Yeah! sound art.
[Cymbol Crash]
OH YEAH!
OH YEAH!!!
OH YEAH!!!!
OH YEAH!!!!!
OH YEAH!!!!!!
[Interlude]
[Cymbol Crash]
OH YEAH!
OH YEAH!!!
OH YEAH!!!!
OH YEAH!!!!!
OH YEAH!!!!!!
[Interlude]
[Cymbol Crash]
OH YEAH!
OH YEAH!!!
OH YEAH!!!!
OH YEAH!!!!!
OH YEAH!!!!!!
[Interlude]
[Cymbol Crash]
OH YEAH!
OH YEAH!!!
OH YEAH!!!!
OH YEAH!...
I am curious how well I do on lyrics that I'm only giving myself 10 minutes to write from nothing
If you extend from silence, it often gives very short output. Try moving the extension point to an earlier time
Are you extending at the very end of the song, when the song has nearly finished? If the songs sounds like it's wrapping up when you extend it, then the extension will also wrap up
I don't click any links but from suno , so no YouTube links here for me. Sorry.
the song ends. and the finished song is 2 minutes for example. When I EXTEND this 2 minute song, it creates a separate song with a 6-7 second extension.
Oh Yeah! sound art. song. Listen and make your own with Suno.
Right 👍 İ have to start from 29 seconds gör extand
any mention of Kool Aid will only remind me of BMTH
If every song is only 10 minutes to write, yea, maybe the music is too entrancing to critically see your lyrics. But hey, some great songs take 10 minutes, or 10 days, others 10 months to write the first draft. So who knows.
Generally, we see our newborns as perfect, but hang out with then awhile until they are teens and we find lots of problems we knew we should have been attentive to earlier on.
Assume your first drafts and couple of rewrites are not your final song. Or rather, always write three versions of your same song and I'll guarantee you'll find one is better than the other two.
it's because the song has ended and the AI is following the trajectory of the song. If the song is over, or fading out, the AI will follow that trajectory to end the song. Try extending from a point where the music is still going strong
You've missed the point of my 10 minute limitation. t was an experiment
I spend long on my meaningful lyrics
it was a personal challenge
I get it, for sure. Got one of those in a text file myself
I see nothing about the lyrics as perfect, they're random rhyme schemed, nothing else
I see, like ex:20th second 👍
Up 'til now, I'd been starting from a base given to me by a GPT response, and punching-up lyrics... now I want to try writing my own from scratch
My most important lyrics will never ever touch Suno cause I'll sing them myself into samples, loops, melodics etc
it's fun to write lyrics
or cathartic, depending on the lyrics
yeah, I'm with you on that. My best bangers aren't Public here. I take 'em to HonkFM
Ahhh... sorry I missed the point. You seem like you have a plan in action and are doing well. Great! There's no one best way to write. That's certain.
Suno definitely helps.
all Suno does for me is samples, loops and rhythmic inspiration
In the end I want to sing my real lyrics myself
but Suno will def help me get stems
Suno is just another crayon colour we can add to our box. Yup.
@safe hazel Can İ REMİX mu finished track on Suno ?
@polar spade @safe hazel Can İ REMİX mu finished track on Suno ?
nope
No. We can't upload stems and remix in suno. It's not (yet) designed for that. It's music generation than it is for editing.
There are much too many DAW style editors out there, free or otherwise. So the assumption is that doesn't need re-inventing -- as it relates to a new concept of writing songs from scratch with AI.
Cakewalk is a free DAW in it's legacy form
I think I still have my cakewalk CD somewhere.
@polar spade https://suno.com/song/8b3e20f7-6f87-45b1-a61c-e3edfddb409e if you want lyrics I find more meaningful
Metal, Metalcore, Nu-Metal, Hard Rock, Emocore, Glitchcore, Drop D Scale, Instrumental Gothic Melody song. Listen and make your own with Suno.
https://suno.com/song/d116bb2d-a459-4e8b-b219-0b2384efde2c or this version
Dark, Sad, Emo, Industrial Rock, Emocore, Emotional song. Listen and make your own with Suno.
The lyrics while I haven't spent that much time on them, are close to my life
the second version is better though, the 2 minute one
@polar spade thnx...How can I describe the sound effects and instrument sounds I want to Suno? For example, if I say apply a sound effect to the 20th second, will Suno do this?
you can't do that
PROMPTS FOR AI DANCE MUSIC @mista.lewys @normalgoodz / Lewys C. G. Evans Hello and welcome! Thanks for opening this ebook and your mind to the future of music creation using text. With this book and a creative spirit, let your imaginations come alive with artistic and specific prompting ...
you can't dictate an exact time that you want something to occur
you can do [Verse, SFX] I guess?
I mean I can specify to do something in a metaprompt
@polar spade what you think of the second version ?
the only one I published
It is not done in 1 step, to have a sound or style or instrument drop in at a certain point.
- You can write lyrics and include metatags, prompts in brackets which can nudge suno to include something at that point such as
May had a little lamb
It's fleece was white as snow
[Laughter]
[Solo: cello]
And everywhere that Mary went
The lamb was sure to go
Prompts in that manner could have it laugh and play a cello solo in the middle. It is not guaranteed, but has been helpful.
Wow thanks, I'm bookmarking this.
np
it's all just links from the wiki
I can't promise they will work
Ok İ see👍
I've had [gunshot] work in songs. Oddly [gunshots] didn't.
what would [Gunfight] do, by chance?
I would just add them in the DAW with a bullet stack
@polar spade I mean if the lyrics are shit I would prefer to hear that ;D
gotta do an extension on it and work on some shifting into heavy metal
Were they good sounding gunshots or a percussive sound that sounds close. I've gone through gens listing "gunshot" in the prompts and so far just hear close sounding percussive or orchestral hits. Cool if you got actual sound effects.
or what about [shootout]?
Try them and get back to me. Lol. "My precious credits, mine mine mine"
[Bullet Stack Crash+Kick+Percussive bass} maybe 😛
I dunno
time to see if I can do a rap song without having it end after 30 seconds cause of impatient ai
rap doesn't do that well with following regular verse, chorus structure 😛
How to get correct "pronounciation" of specific word in my language? I am trying to generate indonesian christian song which contains "Allah" (meaning:God), but SUNO use the majority moslem friend pronounciation in arabic "Auloh" instead of how it should be. Please help.
Is there a synonym somewhere?
AllAah maybe
or well maybe not
maybe it says "all" then
but Aa is a drawn out A and A is a quick, when phonetically written
in english, sure
but what about Indonesian?
was wondering if there was a synonym that was more specific to the Christian God
'cause english uses both, with differing subtexts (of course, english is the language that follows other tongues down dark alleyways, mugs them for their grammar and rifles through their pockets for spare vocabulary)
Well does the LLM recognize indonesian?
https://suno.com/song/b6a98940-fd59-4bf4-afe8-d885998ece12 did I manage to hit the tempo on this cyberpunkish electro-rap?
100 Beats Per Minute, Rap, Hip Hop, Electronic Music, Cyberpunk, Tempo song. Listen and make your own with Suno.
@polar spade @cunning mulch this one and the other one I linked earlier are when I work on lyrics more than 10 minutes 😛
Thoughts on it though?
oh, I haven't listened yet
and I wouldn't be qualified to be able to tell if you hit your tempo :/
Is there any way at all to cut a song off after a certain amount of time? Cause it goes quiet with 30 seconds left
well the qualification is if the rap sounds right enough with the electronic
you can Extend in v3 with just [End] in all the prompt fields. That will, however, mark the 'Whole Song' as being v3
so it does end
bah
I can't past the lyrics
[Rap Verse]
Tomorrow is a revision of the decisions
Envisioned from a position of 20/20 vision
Perfect hindsight of a regression
From yesterdays obsession
Never learn my lesson
This is my confession
I forgot to pay attention
To my own increasing aggression[Beat Drop, Breakdown]
[Rap Verse]
To trap the monster from my past
Disasters left to last
From every darkest recipe
For all the misery
The victory was all a trickery
To fool me into complacency
For the mentality that destroyed my vitality[Rap Verse]
Medicated disintegration
My obsession, no hesitation
Always resignation
Autopilot and casual obsession
Have you learned your f******g lesson?[Interlude, Rising Tempo]
[Rap Bridge]
I lie still and wonder why I left inside
Wire me, desire me
Catch me, release me and send me
The frozen bullet and the gun[Rap Verse]
System raging
I'm engaging with the aging
Of my sweet sweet life
Desperate for time
Consider me a wasted opportunity
Just let me be cause I foresee
That life without me will be free[Ending] [Refrain]
there
it has ending and refrain
yes, but if you Extend from where the silence begins, v3 should cut it off with a 1-second track
Oh I see
but, when you do Get Whole Song from that final Extension, it will be tagged as done in v3
Doesn't matter to me
fair
I hope my rhyme scheme was good enough 😛
I think I'm gonna have to throw myself into the world of mastering soon too
I don't get technical on bars or you know scholarly and stuff like some
fair
so I have no idea if I hit the syllable count
I know some music theory, from playing wround with MIDIs and trackers
well, you should be able to count syllables
eh I cba to learn music theory for producing or where I will later learn to play guitar,
I can start counting syllables but I don't structure my whole songs lyrics based on them, actively
that's fine
I go with my gut feeling on what "feels" right
Eh....I mean Eminem never did count it he just went with pronouncing words with vowel bends
yes
and on beats
but Suno is based on a whole shit-ton of rules
it is but it doesn't seem to follow half of them
so, if you decide you absolutely have to have a line that's much shorter in count, you have to be prepared to guide Suno into some type of pattern you want
The problem I hit with this song is that the tempo is right but the electronic instrumental is wrong.
wrong how?
It's wrong
Just wrong
I didn't want that simple techno flow
but I did want the vocalist speed
I'm more interested in figuring things like how many bars Suno generally dedicates to lyrics per line
You need to align it with the correct bpm
https://suno.com/song/394557ec-766e-447a-906d-a0507f386fbd here's another version of it
Hip Hop, Rap, Spitting Bars, Rhyme Schemes, Electro Beat Music, Cyberpunk song. Listen and make your own with Suno.
it has less bars too
like, if I feed it a line of 9-16 syllables, is it going to use 3-4 measures at 4/X (1-2 at 8/X), or what?
I dunno, I just go with the flow and it seems to work out
well, most music today is based on 4s
but maybe I'm that special kind of artist that's so messed up in the head, so mentally screwed up that I can do that intuitively while with the rest of my life...I act like the kind of guy to jug gasoline and eat poison ivy
(4 lines per verse, 4/4 time signature, etc.)
Eh I listen to a lot of music, where rap is more fused with other genres and they don't follow it much
there's also the 3/4 and 6/8
...you can see why it wasn't considered 'real music' for some time
yes, there's those waltz-like signatures
they come with breaks where you pause mid-word
but those odd-balls are the exception for a reason
and are generally used best as a spice to a more standard 4-base
Push the trigger while I'm rigg-ing
The sin-gle forgotten sinn-er while
Swim-ing in the deep-er sin eater
Part where I'm reach-ing the pieces
etc etc
bit I just went intuitively there, I have no idea how to rap
I just thought it sounded right
like, go ahead... try rapping in 11/8
Takes time to get into the speed for English for me. I can speak really really really really fast in Swedish
but not English
so rhyme schemes etc will also take a bit to get used to if I want to do them
obviously
but I prefer metal, grindcore, deathcore, hyperpop, futurepop, synthwave, cyberpunk, rock, glitchcore etc
unless you're native-fluent, you're gonna take longer in a second language
eh not for singing though
I can speak English without an accent just fine
but rapping I'm not used to at all
I never did it in my life before
wrote my first "rap lyrics" 3 weeks ago, then barely anything else
Swedish contains the entire English phonemic library?
I mean, you can be exceptionally-gifted in reproducing foreign phonetics
I can do Welsh, English, Scottish, Londoner, Indian, American, Southern American etc but not New Zealand, Australia or South Africa
I mean Indian accent, notl anguage
I can't do the language at all
and obviously I'm not gonna go around doing an Indian accent with people who don't know me or they will cry racist ^^
and I don't want to offend or upset people with no reason
I find it harder to do a Swedish accent than to not do it, my childhood friend is the same way. It could be because we've spoken English since we were 9
that would do it
higher neural plasticity in childhood means easier language acquisition and retention
that doesnt explain the other accents but I could just be ok at doing them and that's it
thanks to my medicine I still have high plasticity
(Modafinil)
Yes but modafinil increases neuroglial activity and plasticity by a lot and yes we're off topic, way off
but anyway...If I'm gonna rap I will probably do that half-singing style of rap
I think that might work better with metalcore etc, when I can't reproduce what the AI voice does
speaking of which I need to generate something new...Give me a challenge @cunning mulch
try something technically difficult that is perfect for this channel
...fine. you want an experiment challenge?
yes
what do you mean?
any songs where the verses might have a clear narrative structure to them
maybe not quite no chorus
but where repeating earlier verse content would break the thing
but you usually wouldn't repeat a verse
3.5 is notorious for wanting to try and repeat lyrical content way early, which can be infuriating if you have plenty of lyrics left to have processed
I haven't had a chance to test, but would something like 'narrative structure' in the style prompt tell it "hey, keep it straight"
I have songs popping up in my library that I didn’t create using my credits 😓
are there credits missing?
doubt it
but look at something like Mississippi Squirrel Revival... that might be spoken verses, but tell me how you can repeat verse lines in that outside of original context and have it work
https://suno.com/song/5a19b5d2-ef5c-4545-b77e-df30c694fa5a hows this for a story? 😛 Took a lyric I've worked on awhile back for awhile
High Tempo, Dark and Melancholy, Hypercore, Glitchcore, Metal, Futurepop, Synthwave song. Listen and make your own with Suno.
then added instructions and it followed them
I need to go in and see what kind of rhythm control I can get with lyrics using punctuation
A lot more
how much?
Well it knows when to stop instead of continuing a sentence
if it continues it might become too long and miss the tempo
right, but I'm talking about general rest note length for each symbol
yeah
I noticed that. Keep 'em short and high-signal
gives you more room for lyrics too
see, I'm at that mid-point where I have an idea what I want, but not the verbiage to express it the way Suno wants
but at the same time, there is a "surprise me" factor I want Suno to tap into
well I started Suno today and this is like the 4th song
so it's not that complex
just write simple instructions and try to keep the cadence going
then maybe I really don't know WTF I'm doing
but I literally dumped over 4K credits into this damned song...
Dude....
That means you should have stopped long long ago and waited a few days and then listened again and tried to hear what was missing
home page looking good on firefox now, glad it was fixed so quickly
Just keep aware of your own cognitive biases, and try to learn something from each generation -- if you make small enough adjustments you'll have an easier time telling what worked and what didn't. 💪
Even if you try something and it doesn't work out the way you want -- it's only a "wasted generation" if you're framing it as "Suno not listening to you", requiring you to "know the right words to use". It's a literally a new form of instrument, and it sounds like you're making good steps familiarizing yourself with it.
I genuinely believe the more you start to grok Suno, the more you realize how hard it is to give an exact answer to "how do I make suno do XYZ", and you try to reframe how you try to coax sounds out it.
sure, I'm more concerned about getting the instruments sampled out though
but if I have the right tempo going it will work
well, I ran into suno being completely unwilling to not repeat lines 7 and 8 of the first Verse before the first chorus
no matter what types of adjustments I tried to make
so, call it about 1K on just that alone
but...I dunno why you're telling me like I'm the one fighting this
probably another 1K on trying to get an Extension to proceed directly to the chorus (which it finally did, but in 'v1', it caused an audio hiccup because I cut too early... yet otherwise, even the extensions would start repeating)
I'm not telling you. 🙂
You're right, I'm not part of this conversation and I was trying to offer supportive agreeing statements of encouragement, but it might not have been my place to do so.
that wasnt what I meant
🍻 💙
and then, finally, later on, I probably dropped another couple hundred credits while being affected by a bug that caused suno to flat-out ignore any lyrics tweaks in the Input field
I meant that @cunning mulch is the one I'm trying to help out from my newbie perspective, so I didn't know why you were advicing me @charred merlin
at least that last problem comes with a work-around
he doesn't seem to like me much for some reason
Not ungrateful and not telling you not to help 😛
Not the case in the slightest dude ❤️
ah I love hypercore, glitchcore etc so much
It was generalized advice for any new user -- I was adding a section at the end about how you seemed like you were on the right track, despite how early into your experience you were.
I genuinely feel like there's a big degree of luck in the process, but the fact that you're starting small is a GREAT way to build a "english-to-suno mental dictionary".
well, when you twice repeat a falsehood, that's going to be understandable to think
ah okay, gotcha
I never said that you said 'it doesn't happen', I said that you said you couldn't reproduce it
but that conversation led me to question whether the test itself was flawed
well @cunning mulch may I try to make your song with one or two shots if you have the lyrics and if youcan explain to me what you want out of it?
I'll delete it after I DL it for you
and that's why I'm now exploring the possibility it's linked to using combined songs as a base, rather than a fresh-start 'Part 1'
since that's been a common factor in everyone who's been bitten by that bug
I mentioned it earlier, I spent a lot of my own personal free time trying to help you.
When we weren't able to understand each other, you made a comment that I considered pretty rude and disrespectful of my efforts, and seemed to place the fault of the misunderstanding solely on me.
I don't dislike you, or I wouldn't respond, but it seems that we DO have communication issues. I apologize if you feel like I've been rude to you, I simply tried to express that I'm not sure that we tackle problems in the same way.
well, I think sometimes your methodology has flaws
#📚┃suno-school message This wasn't meant because I dislike you.
<3! Wonderful!
That's what forces science forward! If you disagree with my methods, you devise your own! 😄
The great thing about science is that it happens regardless if people agree with eachothers methods or not -- the only caveat is that unless someone else is able to get the same results (peer review), it's more of a curiosity than anything that stands on its own.
well, tell you what... if I can experiment on this kind of stuff not on my own dime, I'd be happy to start working on it
but I really don't feel like dropping $40-50 a month to be a bug tester
that's kinda the reverse of that whole relationship there
100%, I don't expect you to do this.
Like I said from the beginning, I'm willing to spend my own credits for experimental purposes, but that doesn't mean that I spend my credits without a proper testing methodology.
dude I offered to give it a go or two if you gave me lyrics + explained what you wanted out of it
that would be on my dime
it's fine, man. My lyrics are a bit controversial
As it turns out, some people who are actively helpful to the community are given credits for testing purposes 👀 The expectation isn't that people "waste credits", but try develop your own understanding , or to interact with helpful community members & the staff-elected guides (the people who the staff feel are helpful)
sure but I'm just really f'd in the head, in all kinds of ways...I even have lyrics where I mention enjoying it and wanting to do it again
and rap artists more or less glorify drugs
I talk about them like a noose
and worse, those use them as bragging rights
a very lovely noose
and thats what makes it so dangerous cause you're perfectly happy to hang yourself from it
only realizing you're dying when it's too late
no a noose made of hope and joy
to escape with another thorn through my neck
To pierce my throat a lovely touch
A gentle knife it cuts
Cuts my veins and bleed me dry with all the misery inside.🎶
or something
if I had to make something up on the fly
but the same general idea
the velvet feels so good around your neck, right up until the drop
it feels good after the drop, it's not until you're suffocating that you realize you f'd up
I will never ever even hint at the suggestion someone take drugs. It's a terrible terrible decision
eh... I mean, I'm a cannabis user
I keep it there though
...yes, the Schedule I, no-medical-use, cannabis vs. the Schedule III, available-at-your-local-pharmacy opiods...
but once again if you would like me to try and get your song working, I'm willing to use one or two shots of my credits
it's fine. I think I'm gonna try to use v1 as a way to learn mastering
Oxycodone, fentanyl, methadone, morphine, amphetamine, cocaine, opium etc....All very very bad choices
fentanyl for sure
if you wanna learn mastering, try to stem separate them entirely
that's part of the process I need to learn
literally starting from 0 on that front
I think FL Studio producer edition comes with a plugin for stem separation
so you can take any song and it will pick out drums, guitar, bass, vocals, percussions, synth
and separate them
but I know I want to split stems, record my glass-shattering take on the vocals, and then get them AI-replaced
I don't see how you will get your voice replaced by ai voice in suno
maybe on elevenlabs
correct
I'm not looking at suno for any of that
I see suno as... kind of a 'jam band' in a black box. You pass them your lyrics, they bounce ideas around and come back with a take
Suno is great for rhythm, cadence, guitar riffs and such to create 808s
If you get something you like, you can use Extends to work on it and have them take it from a specific spot
if I get something I really like, I'm cutting it up and using the samples
..but really, I'd want the ability to say "Hey, that was great. I've punched up lyrics a bit. Let's do that again with the new ones, and maybe work in an intro"
I hated the idea of throwing out a banger to have to start from scratch
I guess we have different thoughts about generative AI
we do
I don't consider anything the generative AI does as being my creation as if I should be proud of it. The only thing I will be proud of are lyrics
my goal is to produce full songs with it, and then master them
but if I take the samples and such and use them to create 808s and synth samples
that's different
but for most songs, lyrics are the biggest part
I'm lucky to be a great writer, for lyrics, poems, fiction, stories etc. Not so much non-fiction
at least the ones I enjoy the most
or non lyrics
but generative AI stuff, I guess I'm used to how it's not really me doing it. I mean I created models for stable diffusion
and even after making the models, I was still like "meh it's not me making it, not really"
but that's me
I'm very harsh on myself
as am I
I can't see how people put out so many songs in just a few days, unless they're just accepting what comes out
ya, anyone who has ever learned an instrument will agree
unlike many "synth artists" I don't consider the use of samples or loops to be cheap though
but it's about how you use it
im doing exactly that, mass for the content filler, then i work on some gems just for myself
just playing full samples and loops and doing nothing with it at all is cheap
but taking it and manipulating it to fit you and using it to make your song, awesome
see, my previous 'creation' experience was with tracking. I liked making covers of older gaming music in the format
I'm waiting to take acoustic guitar lessons soon. All I want to learn is the basics which is technique, best way to hold the guitar, muscle memory and reading tabs
after that I can learn the rest on my own
so, i suppose I was working more in transcription
instead of making covers, make reimaginings
yeah?
If you want an example of what reimagining something is. Check Falling in Reverse - Last Resort
their take on Last Resort is hauntingly beautiful
instead of Papa Roach's angst
Gotcha
(also Jacobi approached FIR about reimagining it and he loves it. It made him break down and cry)
for me, it was about porting skill. How well could I take the results of what was programmed into a pulse generator, and replicate it in a vastly different environment
you gotta be ready for their Last Resort being very VEEEEEEEEEEEEEEEEERY different and rather than angst it's about desperate cry for help, the haunting loneliness and sadness
Sure. I like beautiful things
Reimagings is cool because it adds an element of artistry in the very act of making it
yep
so, with Suno, the skill I'm trying to flex is AI herding
and then, maybe music production, since I'd be mastering
I'm more tech-minded than creative
/chirp
Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com
and if you want to see the sharp contrast to Radke's range, just check his other songs like Carry On, Losing My Life, Losing My Mind, Drugs, Zombie, Voices In My Head, Popular Monster, Watch The World Burn, Ronald, All My Life, I'm A Vampire (Reimagined), The Drug In Me Is You (Reimagined)
or, better stated, my creativity is expressed more in revising and technical areas, than in others
basically, I tend to play best in other peoples' sandboxes
I mean it goes from Synthwave to Futurepop to Synthrock to Metal, to Nu-Metal, to Rap Metalcore to Deathcore to Country to Ballad, to Piano, to Orchestral
if that's not range I dunno what is :p
but yeah "Ronald" is heavy stuff
I relate much more to Bring Me The Horizon lyrics though
but as far as music goes... I'm in no way mainstream
@chrome scarab did you manage to catch their Last Resort?
Listening right now
I don't care about mainstream or not mainstream, if it's good, it's good
the bulk of my phone's MP3s are of gaming soundtracks
stopping liking a band cause they get famous is a bit snobbish imo
not saying thats what you did
hello, what should i do if i want to change one sentence in the middle of a song?
as far as the rest... parodies now probably outweigh 'mainstream' songs
either toss it out or wait for inpainting
@oak jungle You can extend it starting from the middle
there's no wy to fix something in the middle of something you like while keeping the rest
😮
you'll lose what you had up until the point you re-worked from :/
I'm gonna get some vocal coaching from a eurovision participant. She represented Iceland one year
but now she lives in Sweden
#1270928841614823546 message video tutorial on how to extend from an earlier segment ❤️
Extension is not a destructive process, and will leave your songs in tact. It only creates a stitched version if you "get whole song", but again, this leaves the originals in tact, allowing you to make as many adjustments you'd like to find the best extension point.
she's also got CVT certified training from CVT academy in Copenhagen
it's frustrating, but I got a couple songs like that that I'm eagerly awaiting inpainting for
Like I said, just do extend from the middle
then maybe adding in the audio with the last part that you lost
so it can try to lead up to that
right. The way he stated it sounded like he likes the rest of it, except the line in the middle... the unfortunate reality is that in order to change said line, it'll be all new after that new line
yes that's what i want,
then use extend
i want to keep everything, and edit one lyric
oh
then nevermind you can't
or you can copy the song, extend it for one lyric and then use a DAW
Not currently available with the tools available -- you can edit it externally SUPER easily, but we're giving you tools on how to edit it within Suno.
inpainting confirmed?
hmmm
Not even available for alpha testers, as there's literally no news on the new model. But the idea of inpainting was being discussed even before I was here.
oh. your wording sounded like it was close
So it's in the pipeline, but not worth putting off learning effective use of current tools.
which means that he also needs to be made aware of the downsides of said current tools
Yeah, it would be GREAT to change just a very specific section, but I find the best success by "locking in" earlier sections and building off those -- if you REALLY like a piece from a later section, there's ways to save it, with DAWs and creative song production methods.
However, after making massive external changes , the songs can't live on suno.com unless they're 6-60s for reupload, so my "official songs" live elsewhere, and my "fun stuff" stays on Suno
re-generating from a certain point will wind up causing the song to change
and there's just no way around that at present in Suno itself
yeh, if you cant afford a DAW just get Cakewalk Legacy by Bandlabs
it's free and very versatile
https://legacy.cakewalk.com/ this one
Cakewalk develops computer software for recording and making music. Our products include award-winning digital audio workstations for PC, fully-integrated music making software and recording hardware, and innovative soft-synth virtual instruments for PC and Mac.
no I don't think so
ok i'll try i'm a real newb in this things ^^ i'm just making a funny song for a friend who has cancer, the song is about her and me 🙂 but wait just to make sure, i'll try to explain better: i have a song with 2 voices, first voice sings the whole song, the second voice, only says one sentence, then the first voice sings again, and i want to change the lyrics from that second voice
I'm oldschool and love audacity for FOSS free software 🙂 got a redesign recently
but you could always buy FL Studio Producer Edition
well CakeWalk being free
I guess that works then 😛
get cakewalk and use splitter
free DAW, got plugins for other stuff too
well, I was thinking finer extractions
if you want that you gotta pay
or become a pirate
Im not gonna help you pirate stuff
I'll keep it real. I'm not looking for total professional level
especially not when FL Studio Producer Edition is like 200 dollars and on black friday is down at like 100-140
and that means lifetime updates
every new numbered version free
I just want to get to not "obviously AI"
then you gptta pay a little dude
FL Studio is the cheapest lifetime update DAW there is
Studio One is second one
BitWig third
but all DAWs can do the same damn thing they just have different workflows
so, as long as I can clean up the sound, I don't care if I'm working with a combined instrumental, or individual instrument tracks
then get cakewalk, it's free and you can do anything any other DAW can do assuming you have the plugins
cakewalk legacy that is
Suno's splitter is... I have one 'vocal' track that contains a random guitar note in it
and the less said about trying to mix the vocal back in with the instrumental in Wavosaur...
god damn it woody, don't get a broner
there's a few other free ones but cakewalk is probably the best free one
FL Studio is probably the best value one
logic pro is MacOS only
pro tools is a fortune
I'm leaning towards Cakewalk
in my budget and probably good enough to do what I need to with it
outdated how?
or find some site online that lets you do it
unless you talk about piracy, there isn't one
I'm trying to decide between FL Studio and Studio One for my DAW
I found a couple that talked about splitting stems and seemed to do it well enough
I mean sure there's cubase but along with Pro Tools that's the most expensive
but I was talking other tools to help you learn about a given track
I don't think you'll find something free that'syou know, better than DAWs'
yeah won't happen, unless it costs money or you pay a subscription for access to VSTs
etc
but this one was useful enough that I even bookmarked it while I was trying to gather information on my previous version to try and recapture some of it: https://www.mazmazika.com/chordanalyzer
Mazmazika. The Free Music Oasis. Vocal Remover using modern AI. Elegant Virtual Piano. Artists with 1 Billion+ Youtube views.
you realize you can click "separate stems" on Suno right?
and it will split vocals from instrumentals
yes
Suno's splitter is... I have one 'vocal' track that contains a random guitar note in it
and you know cakewalk comes with a splitter
I'm looking at the chord analyzer
was for trying to get Suno to come up with something close to what I had in v1
chord analyzer? you mean key?
I uploaded my 'completed' version of the song:
"(C, Em, D, B, E, Bm)
are the main chords of this song.
Explore the chords of the song in real-time,[...]"
and there's a display at the bottom where you can play it back, and it shows the chords, and for how many beats
you can also just move the chord too if you want to
every DAW has BPM, metronomes, chord selections etc
most DAWs have key finder and chord finders
brb
if you want a good reliable DAW though and youre more into synth/midi made sounds, virtual instruments and electronic music. Then FL Studio is probably the best option for the workflow and for their FL Studio Cloud
which features a lot of very expensive plugins, vsti's, samplers etc for 6 bucks a month
if you can wait until Black Friday to buy a DAW though, FL Studio tends to go on sale then
i tried with the extend, but yeah, the song quickly changes after some time, which makes the song completely weird... how do i edit just that one sentence externally super easily? 😄
I'm trying hard to decide between FL Studio, Studio One and BitWig
@oak jungle , like I said, you can take one version and extend it for one change and then have the other wav downloaded int its raw form and just paste the change into it
And then me who upgraded to the all plugins version a few years ago and don’t regret it
and upload to soundcloud
if you can afford it sure that's great
I can't, not even close. For me it's likely Studio One 6 or FL Studio Producers
over here, we're trying to keep it to a far more modest budget
I already did it years ago when I had producer edition and then it was on sale
Studio One 6 Hybrid+ sub is great
yeah fl studio all plugins edition is still too much
The "super easy" method is to place an extension point before the wrong lyrics, and regenerate the rest of the song.
it also doesn't technically feature all plugins since new ones show up on cloud dont they? Or maybe it updates with them
@oak jungle the project management triangle helps in decisions like this.
If none of the sides appeal to you, you can go hunting for the unicorns that exist in the tiny triangle in the center
I think all plugins edition gets them
https://www.image-line.com/fl-cloud here's the FL cloud sub and what it contains
But the rest of on the fl cloud sub
Studio One is a workflow I need to check out cause the only thing separating DAWs is the workflow
and included plugins of course
but other than that, it's workflow
anyway, harber, about that whole 'credits for testing' bit... I wouldn't mind, if it's simple as getting reimbursed for the credits used in obvious testing
I didn't mean to suggest anyone could apply for credits -- specific users get credits because they've proven themselves useful and helpful in many ways.
No one is asking you to.
sorry i don't think i understand :x isn't that what i just did? i clicked extend, changed the first lyrics, but then the thing is that the song changes after some time, and it's not even the same melody :x
https://www.presonus.com/en/studio-one-plus-hybrid.html here is the studio one sub
Studio One+ Hybrid is an annual subscription to Studio One+ with a Studio One Pro perpetual software license included. Join today to gain access to the ultimate drag-and-drop digital audio workstation, all add-ons, content, and extras.
but we explained this to you....You will need a DAW or a audio tool to combine em
You asked for a "super easy" method, and I'm trying to figure out which side of the project management triangle you're favouring.
If you want "super easy" and "high quality", it will cost money, like using external tools, or getting the help of an audio engineer
If you want "super easy and fast", use the suggestions I'm giving.
If you want super easy, fast, and low cost / free, you need to hunt for unicorns (like bandlab, they're a unicorn in my mind, but not for this specific usecase)
simply cut the part out that you changed and paste it into the full song
or get cakewalk ;P
Why I like fl is one time purchase and free updates and it’s just how I got used to the workflow (even though my stuff is terrible in comparison to sunos)
oh ok sorry guys i have no clue what a daw is haha
Digital audio workspace
external tool for working on songs
Digital Audio Workstation, basically music production software
workstation* 😛
studio one hybrid+ is great for its sub with plugins, vst's, vsti and yearly perpetual licenses at the last update you get on your yearly sub
if you like its workflow better
#📢┃announcements message If you want to get involved, nedi sends out notifications for surveys.
#😜┃suno-memes message some people meme about it, but this is literally how I got involved with the team. My detailed and helpful feedback must have struck a chord, because they asked me to provide alpha feedback on stuff.
It's not a matter of someone going to them and saying "give me credits because I need to learn"
I've seen 'em
FL Studio is still the most value though
if the workflow works for you
Pro Tools is the one all...pro's use lol
not that it means it's better
doesn't
That’s just how I started back at FL 12, and stick with since, if other daws have better workflows for, that’s great
I'm gonna test fl studio and then studio one and then bitwig
No, it would be more of a case of "I spent X credits in testing [problem] that users have been experiencing to find a reliable reproduction. I would like those reimbursed"
ableton is too fucsed on instruments, like for a band
I think API for lyrics returned this "error" but Suno was like "Of course! those are the perfect lyrics!". I asked for a choir song with no vocals xD
choir with no vocals would be chanting
Cool! I'll try it right away ^^ thx
You can make songs in the fl demo, but can’t reopen them unless you buy FL
You may've missed it, but I said that it would be more of a case of "I spent X credits on testing [problem] that users have been experiencing to find a reliable reproduction. I would like those reimbursed" than anything else. Not asking for advance... just getting back what I used
aaah
and not like it'd be hard for them to verify, either. A bunch of tracks with styles and lyrics way outside my usual personal range, with obvious testing names... yeah, those would be what I'd be putting in for
but why do you think you deserve reimbursement for allowing yourself to let things get out of hand?
not that you cant try
doesn't hurt to try
but I think you should slow down a bit
I need to start making samples. Is v1 best for samples or v2?
or v2 even
v2, v3, v3.5
trying v3 I suppose hmm
bah drums
no negative prompts either
Something must be wrong with my prompts now @cunning mulch , 8th generation in a row, where it repeats the Verse 1, 3-4 times, how did you design it again?
are you using punctuation?
I'm experimenting with punctuations now, i've only used commas and new line seperations for awhile
Otherwise you can leave a line unfinished and it can loop back around
sometimes it does it either way but, don't name chorus or verses
Hmm, still jumps back
put simplistic information within []
I've used Suno now for so long, and still cannot figure out why it does this, over and over, and then magically "it doesn't"
https://suno.com/song/5a19b5d2-ef5c-4545-b77e-df30c694fa5a I mean here's my latest
High Tempo, Dark and Melancholy, Hypercore, Glitchcore, Metal, Futurepop, Synthwave song. Listen and make your own with Suno.
you can take some.....suggestions from my structure I suppose
You're just using more lines then me, is all
[Intro]
Demons, Angels, Saints or Sinners
I don’t care.
They’re all screaming my ear.
Like a virus in my brain, it’s making me so ashamed.[Industrial metalcore guitar solo]
[Verse 1]
different voices of different kins
I don’t care which one it is
Cause all I hear is tomorrow and yesterday
But I never hear about today
How do I survive this day?
Somebody tell me.
I’m too afraid[Chorus]
I load the gun once again.
Push the pill and nothing said.
A chemical virus that renders me happy and blind.
Each time I pull the trigger I wake up and taste reality again.
Pleasure is cutting up my brain and I'm going insane.[Instrumental Orchestral, Electric Guitar Metal Power Chord]
[Verse 2]
I’m so terrified.
I don’t want to die.
Time for a mind suicide.
I don’t want to feel the fear
That haunted me for years
I can’t run, I’m paralyzed
As they eat me alive
And I’m covered in sin
It’s all beneath my skin
I think it’s gonna win[Chorus]
I load the gun once again.
Push the pill and nothing said.
A chemical virus that renders me happy and blind.
Each time I pull the trigger I wake up and taste reality again.
Pleasure is cutting up my brain and I'm going insane.[Fast tempo hypercore into breakdown]
[Outro, Harsh Deathcore]
Exact execution on the line.
Place my demons on the walls.
Push the trigger so they fall.[Guitar Riff]
A land of violence caught in silence.
I look for my misery in the sand.
As I wander deserts across the land.
I think I might become a mummy.
Or perhaps a demon would be funny?
Time to die alone tonight.
I forgot to end it with [Ending][Refrain] though
cause that stops a song
What do you have going?
Verse 1
4-5 lines
Verse 2
4-5 lines
I'm trying now to put everything within the same format
I have more than 5 lines
I'm having humming included
dont matter much to mine
Yea
It's Verse 2 & 3 that's the culprit, and Verse 2 seems to love to get repeated...
Verse 2
Morning light breaks, keeps me awake
Harbouring regrets that I can't, shake~
Up all night thinking, up all night drinking'
Drowning in dreams, and the thoughts I'm sinking. mmm. m-m
Verse 3
Songs of the broken
Whispers in the night~
Carrying this weight, heavy on my mind~
It’s that same old place, that I can't leave behind?
Forever wanting to return, let my feet touch the ground
It keeps on going back to Verse 2, over and over
Are you using numbers in the verse tags?
I'm not using a [ ] deviator for the Verse category
marking them as [Verse 1], etc.
I used it before, still did the same thing
well, try marking them as [Verse], if nothing else
I'm not having high hopes with this [ ], let me check
Ok, one of them worked now. So strange, it didn't work before
it doesn't quite know what to do if you're not using SOME markers
besides, you can then re-use those markers by themselves if you want repeated lyrics
I find it strange, how I had those [ ] before, didn't work, removed them, didn't work, added the [ ] back, and it worked
Can't get it to generate choir... I've tried "female chanting" but no voices are generated 😦
I think the lesson is that nothing works very consistently, all we can really do is try to bias the AI in our favor a little bit and hope
Yeah
Just to make sure I'm not misunderstanding this, are you asking me for a credit refund?
Not trying to be facetious, I genuinely think I'm misunderstanding the situation, so I want to take a solid beat to make sure we're on the same page ❤️
yeah I dunno, I'm too tired to think critically
https://suno.com/song/ed8d16fd-2084-49b3-8cfa-a743a27e67a1 some darkcore 😛 I think I can improve it
Darkcore, E Minor Scale, Heavy Metal Power Chords, Cybercore song. Listen and make your own with Suno.
Do you like crazy songs?
I might teach you something new if you're interested
I recommend you start using "Style of Music" soundfx
I'm not after random FXs
And then you can add, a multitude of tags associated with soundfx related noises. You can add distortions, breaks, feedbacks, which would enrich the lyrics
Darkcore, E Minor Scale, Heavy Metal Power Chords, Cybercore, soundfx song. Listen and make your own with Suno.
I'm not making actual songs with Suno, I'm finding samples and loops to fit my lyrics
that hide within the generative
that I uncover with the lyrics
I'm all for creativity with this app, that's also my primary usage for it, purely reference materials
I see Suno as a randomizer button with many tracks, poems that I write, and songs.
Gives me another angle of approach
mhm
https://suno.com/song/6d3ffdfc-85e7-40ec-9f40-b99b5f663a08 maybe this works a bit better
I should try to write random lyrics for a superpop with futuresynth
Can Suno not do duets? Or am I not doing something right?
yes he can have a duet look this ? https://suno.com/song/0b7e3039-39a3-433c-a20c-87172d4893ab
man vocal, drum, Tagelharpa, Animal skin drum, rhythmic percussion, Traditional Norse, dark ritual, Traditional Scand song. Listen and make your own with Suno.
2:58 time
or this duet man and girl https://suno.com/song/489e1055-9c25-4965-bfe0-73d4708bf893 0:34 time did you have this in mind?
emo-pop, emotional, emo, pop, romantic, sad, piano, female vocalis song. Listen and make your own with Suno.
Just seeing this on a re-read trying to understand what's happening.
When I say that certain members get recognition for their contributions, I don't mean that we're thanked for experiments and tests, we're thanked for our continued goodfaith efforts to help new users use the platform and achieve the same kind of success that we have -- for instance, paimon is a force of nature when it comes to community support for a few solid months running, and rightfully so they were asked to be a community guide. They made an amazing guide helping users locate missing songs that I still have bookmarked to show to new users who might panic.
To clear any misunderstandings, there's no compensation ever provided for "tests", as it's an entirely novel music-gen tool there aren't any formal documentations because it's still being figured out -- try looking up a list of "what you're allowed to say" to chatgpt, and you'll likely get some helpful examples, and maybe a list of things known NOT to work, but it's SUCH a subjective question that trying to get upset about chatGPT not working the way you want becomes unproductive compared to realizing how much it's changed our entire lives and workflows compared to a couple years ago 👀
To repeat, no credits are reimbursed for generations not being the style you wanted, because you've decided what to input as lyrics/metatags/style prompts. I was only mentioning that when people are genuinely helpful to the community, the staff recognize it.
Also, to avoid any misunderstandings I AM NOT A SUNO TEAM MEMBER.
Please don't interpret this as "me not liking you", I recognize that you're upset with the concepts I'm espousing and not necessarily at me (or I'm going to choose to see it that way)

Okay, first, that song sounds amazing! I wish I could understand the lyrics. Second, how? I have tried different prompts and keep getting voices that are indistinguishable. For instance, I am trying to make a duet between two brothers who are fighting each other over the same girl. But the voice is the same. I tried a man and woman singing about their frustrations over everyone giving them advice on how to raise their child as if they didn't know their own child. I've tried romance, rivalry, it's always one person and I'm dying inside. end rant
I struggle a lot too, but it always helps to credit myself (Duet vocalis) and then I'm just looking for it sometimes it's a long wait
the point I was trying to get at is: If I'm going to be doing serious testing to try and find a bug to report it, I'd want the credits used for that specific testing reimbursed (by the team), so that I'm not paying to bug test
So you're saying patience then? Yuck
afaik, this isn't something we're able to explicitly prompt for, but I just had a great idea 😄
This is gonna be the first Experiment Playlist since the Pool Party reopened! 🎉
#1256301881966198805
well... I did tag for a 'palette' of chords, and it seemed to stay pretty close
of course, I was working with one set of lyrics, so...
Don't worry, they have internally paid bug testers -- any bug tests that the community does is out of the goodness of their hearts because they love the platform and want it to be the best it can be. 💪 
Not saying that's a job title in and of itself, but trust that ALL of the paid staff actively test things and want it to be the best it can before releasing.
If you're passionate about bug testing, and believe that your time and efforts are worth compensation, I highly encourage you to submit an application to Suno at https://jobs.ashbyhq.com/suno, or there's their Talent Community role for people who don't see an open position that suits them, but who still want to express genuine interest in a position, whatever it may be (you get to choose how to best sell yourself in the cover letter, remember you're applying for jobs that haven't been invented yet, so have fun with it!)
https://jobs.ashbyhq.com/suno/c87801d1-4b65-4f1f-9ede-821e7f84ea4c
well, considering it took another 300+ to finally get something I considered acceptable finished, I think I'm gonna have to watch my usage
can't always spring for a reload
I mean, even the fact that I did go paid is a big thing, since it shows I think there's enough value here that it's worth that
Try to keep the prompts simple, is my advice, and the instruction tags in the lyrics too.
i'll try
i'm starting to think its related to the database :/
https://suno.com/song/a1eb0bf9-3f8e-4369-802d-dbc2cf9734c1
for example, i was extrictly specific on this one, but the vocalists sounds like one of those modern rock bands, not from the 80s
80s hard rock song. Listen and make your own with Suno.
😨 extra credits are SUPER expensive because they have "premium rules" in terms of how they expire.
Cost per Credit breakdown in terms of source
$0.00 -- daily 50-cap for free users, or paid members who use all credits
$0.30 -- premier plan
$0.40 -- pro plan
$0.75 -- 4000 credit topup
$0.80 -- all other extra credit topups
You're free to spend your money however you'd like, and I don't want to discourage anyone from expressing their passion for the project, but I sincerely don't want you to spend money angrily.
That said, if you genuinely feel like you're owed compensation for dissatisfactory service, you should absolutely reach out to the support or billing team by email, but there's probably something in the ToS that says that they make no guarantees of the output, and that by simply continuing to use the service, you waive any rights to compensation, yadda yadda not a lawyer, but you do you
you seem to be reading WAAAAAY too much into what I've been saying
Sorry we keep seeming to misunderstand each other 😅
hope your're having a good <t:1724635440:t>
I mean, sure, I'm a bit salty at that particular song using so many credits, but the worst case is I'd just drop back to basic
sometimes I screw up and don't realize it
you know, THAT just happens
BUT, at the same time, I only get so many credits a month (even with the 'at-0 lifeline' now). I can't justify dropping 10% of 'em on explicit testing to zero in on a cause
Yeah, for the first 3 months, all of my tests happened as wild sprints in the last 3 days of my credit-cycles, so that I could 1) get a mass amount of songs to continue from later, and 2) try things without worrying about "wasting" anything because they were about to reset anyway 🙂
I wouldn't mind doing that kind of specific testing. I just don't want to have to come out worse for doing so
Just realizing how many songs from those experiment-sprints ended up on my album one way or another, compared to how few were "calculated" 😛
I mean, sure, if we get close to the 23rd of September and I've got a ton left with no song ideas...
but at the rate past songs have chewed through credits, I'm not holding much hope
now, I have zero clue where you get the idea I want any credits currently used back
oh my god -- it's a blessing in disguise -- and it only really becomes frustratng if you go into it with specific expectations, but that last 3 days when you don't have time to be super critical, go wild -- pick a specific prompt, only use that for 20 songs, and listen to each -- REALLY listen. dont add crazy extras or lyrics, just the style prompts. after you sort of get an expectation of what to get, try adding a single addition and check how they combine.
ratatouille_synethesia_scene.gif
wow... indie... and math-rock.
Now that I've explored so many different flavours of "indie math-rock"... what if I add "dj scratch"?
This is why I find it ridiculously hard to work with 10 flavours at once before I've really taken time to understand each of them in isolation. ❤️
we're coming at it from different ends
It's a pain in the ass, and I'm not advocating that everyone should do the same. Hell no. This is just how I did it.
Beat me to the punch ❤️ 😄
I have some semblance of what I want, and I'm trying to see how best to get there
I have faith in you, dude. You have a lot more passion about it than some I see who just roll "whatever suno makes" and excitedly share their latest banger.
well, TBF, I have a bit of that too
That’s me, but I’ve been recently doing extend after each verse
but I try to also be more discerning about what I pick
Please, if there's only one thing you ever take away from my rambles and boorishness -- Please send me the first complete Suno song that you've fallen in love with.
I know how much work you'll have put into it, and you won't need to say a single word to catch my interest. I'll make a fresh pot of coffee, throw on headphones, and probably never be more interested in hearing someone else's Suno song up until that point.
I won't need the regular "song and dance" I advocate for in showcase / gen-chat posts, because regardless of whether or not you feel that we're friends, I'm hella invested in your suno-success at this point.

htf is it nighttime already. I'm such a space-case. 🤦♂️ I have shit to DO
Please someone yell at me if I'm in chat for the next 6 hours. Like, hurtful things.
See, where I 'come from', as it were, AI music is currently seen derisively as 'slop', and.... I disagree with that take. I see the potential, and I first need to prove that it's a tool like any other
that you need a good human mind making the true judgements on what makes it in, from lyrics, to the takes you get
you know, it's funny... Literally the first time I clicked Generate on my first prompt... I got two bangers that I basically had to pick which was slightly better
Me a lot of the time
In the end, I use both but make different songs
then, I started my second... and I got something I was... okay with, but could have been better. I took it into the tank, punched up the lyrics, did something I thought was neat with the chorus... and then after finishing that version, I started from scratch and style-shifted it
I should really use my own FL audio for the audio bases to get more bangers out
besides... I'm... not the most comfortable sharing my songs here. I have an audience in mind, and the lyrics tend to reflect that
not trying to get run off the server
not concerned about being moderated by the AI by any means
haven't used any vulgar language in my lyrics, and don't plan to. I prefer my barbs dlivered other ways
anyway... I gotta be off for now
See ya later
Is there a way to remix a song in Suno? (Like remixing a song that you’ve made in Suno)
There is not
Depends on what you mean by remix. You can extend from various parts of the songs to get different versions of the same lyrics for example.
Fix the system so people don't have to continually spend credits forever in order to get the song to end properly!
That is a known issue for 3.5 since like the beginning and alternatives has been mentioned many times.
Basically: 3.5 is not great at ending songs. If you want to terminate a song at a certain time point without using an external program, you will need to [End] it with an extension set to 2.0 or 3.0
- extend the piece at the point you want the song to stop.
- Set this termination extension as 2.0 or 3.0
- Set the style as "End, Short" or just simply End. Up to you if you want [End] in the lyrics
Here is a graphical version of the instructions that was pinned in general: #💬┃general-chat message
I have a question regarding the credits, I had 1k+ credits in my accounts and now 50, I don't recall seeing anywhere that credits actually expire. 😦
Found the answer here: https://help.suno.com/en/articles/2524993 sad that i didn't read the small letters 😦
Yes, paid subscriptions still have expiring credits ! Here's how it works: If you're using the Basic (free) plan, you'll note that you get up to ...
I have a question regarding uploading the content generated with Suno. I understand you need a paid subscription for commercial use, but what i want to do is to upload the songs to youtube just for sharing, i don't want to monetize them or something like that. Can i do that with the free tier?
There is no problem sharing on youtube if it is not going to be monetized, "Songs made under the Basic (free) tier are only available for non-commercial use and cannot be monetized. " source: https://help.suno.com/en/articles/2410177
Yes! Well... let's talk about it. If you made your songs while subscribed to a Pro or Premier plan, those songs are covered by a commercial use li...
perfect! thank you! 🙇♂️
There is anyway to check which music was genereted while I had a pro subscription vs the free tier ?
Note that free plans, in addition to not being monetized, needs to credit Suno (mentioned in the FAQ).
This is as simple as just stating something like "Made with Suno" in the description
isn't it already in the video?
i mean i plan to upload them with the original video 🤷♂️
Suno logo is there in the original but it is not unusual for people to make music videos or such
got it, thank you for the clarification!
If you are interested in videos, there are several services out there. Noisee can make 1 min music video clips, Hydra can make singing avatars like you may see in some of 3Daizy videos, and more
AFAIK, the only way atm is by comparing your email receipt of the plan to the date and time of the creation.
You should suggest at #💡┃suggestions-web about having a more visual appearance for the song status. I know you are not the only one asking that question and I do not know why something like it has not been added in yet
Does anyone know a good method for cleaning up the vocal residue that gets left in the instrumental stems?
Is there any way to correct the genre spelling of a customised generated song?
Hi, I'm trying to create a song but when I click on create it says: Couldn't generate that.
Prompt flagged for moderation
There's a word or concept that the ChatGPT detection Suno uses decided was inappropriate -- try rephrasing, or spelling certain words phonetically if it's an honest accidental flag (but don't trick Suno into making inappropriate songs, because I think that's actually a ToS violation and could get your account locked 😨 )
I don't know if this is still being monitored, but there's a Google Forms page the devs made to help try to fix stuff like this, if you genuinely feel like it's a false-positive.
Error: "Couldn't generate that. Suno is for generating original music. Try different lyrics, and use genres and vibes rather than specific artist names." This means your song was blocked and credits refunded.
Songs on Suno.ai are moderated for:
Non-original lyrics (copyright & artist names)
Being overly explicit (in general or in the use of s...
As bubblegum and unicorns as I am with my songs, I finally wrote one that suno flagged. Haha. But I was really just writing English words to replace the babble-scat to see how it would sound. Now I just need to rewrite it to appeal the gods.
writes an instrumental song -- Prompt Flagged for Moderation
ah lovely LLM moderated by ChatGPT
Haven’t tried yet, but there are a lot of stem separators online/in apps/in DAWs. Eg I like this one https://vocalremover.org/
sign up for Moises or Lalal stem splitter. You'll get 4 to 6 stems instead of one instrumental, and the background vocal hiss you're referring to will be relegated to an "other" stem, which when you mix, you can de-emphasize. But because "stems" with these computer-generated songs are not stems, in the strict sense, it's always going to be there.
what do you use to de-emphasize them? eq?
I use Sony Vegas. I pull all the tracks in and then drop the db on that track, usually also upping the primary vocals, drums, bass, and guitar.
(while also using the original song as a backing track (so nothing is really lost), and I adjust this as well.). If you're trying to get rid of the hiss... then, once it's mixed, have it mastered (again, using any online mastering service), which will further clean things up to a point that actually sounds pretty decent.
Sony Vegas works for videos. But if you like DAWS, just for music, you can also use Waveform, which is free and pretty awesome.
IMPORTANT: when using video software to edit music, set the ruler / measures to Beats (measures and beats, I think it is). Otherwise, harder to slice and dice.
Ooooh, excited to check out those 2 -- I always recommend bandlab just because it's a whole web-suite with a ridiculously generous free tier, but I really should do some a/b testing since there's so many companies quickly making a name for themselves.
Hoping to test them out in the next couple days -- the 6-stem separation catches my eye immediately, excited to test out new tools 😄 Anything else to note going into it, like pricing, what they do well vs still need to improve on?
the v1 feature in Suno, yeah. For now 💪
Yes. When it works, it's helpful, but never perfect. There is light bleeding over for instruments in the vocal stem and vocals in the instrumental. So you try to clean them up post separating. And digitally created vocals seem to come off as "instruments" in my opinion, and so are a bit harder to split off. Often I am losing the vocal when it originally blends into the song. The best splits are when the song is just 1 or 2 instruments.
creating an instrumental. the instrumentals are being generated at 40secs, 1min28sec, 1 min, etc... so 1 out of 30+ generations gives me a full 3min track. wasting a lot of credits for these shorts tracks. anyone know how to alleviate short tracks? ... or is this only happening to me.
You get 4 stems for $6 and 10 stems & mastering fro $30-ish. Some guy on youtube showed that Lalal was the best all-purpose stemmer as they have a broader suite of tools. But I didn't see mastering, so I went with Moises for $6 to test then upgraded when it came time to master. I do not have such an Sonar-level ear when it comes to determining the differences between such services, so, for me, it comes down to "good enough" and easy user interface.
The thing to keep in mind is that offering 10 stems is one thing -- getting 10 stems out of a Suno song is quite another, meaning that the ones beyond the 4 stems are often empty or barely register. So unless you're mastering, the additional stems are often useless.
wait they charge that per song?
The extra cost is for the mastering...
do they master it personally?
AI mastering... I don't think it's a bespoke service.
I only have 54 radio-ready songs. So, for me, it's just the one month.
For someone who produces 100s of songs, you're right, a plugin would be better for perpetual usage.
https://www.izotope.com/en/products/ozone.html there's that 😛
AI mastering and manual in a plugin
55 euro
just saying it might be cheaper in the long run to consider plugins
(if you need a daw to run the plugin inside of cakewalk legacy by bandlabs is free)
of course not telling you what to use, but I like to offer otions
options'*
ctrl + z
omg don't lose that amount of work
Woodhouse, stop eating posts from helpful people
Appreciate your insight and candor on moises and lalal! (wait, those are 2/4 of the teletubbies, right?)
Stepping away from discord for the day, but I really appreciate getting hands-on info about different tools, it really helps knowing what kind of tools standout in peoples minds, especially if they're a user-friendly, easy to learn/room to grow sort of thing
funny, you're right! ha.
this changed recently if youre talking about moises - you now pick which specific stems you want from it up to 5 stems (and then anything not included in those becomes a sixth 'Other' stem). it seems to do a LOT better than it did before when the stem options were 'pre-selected' from a list.
@stiff sable!
hey @charred merlin !!! i've just been doing a lot of lurking recently, busy busy 😓 but still around lol
Same here, lots going on, but still around -- it's been a while since I've seen your name pop up in chat, so I didn't care what chatroom we're in, I missed seeing you around man
Hoping you get to enjoy the fruits of your labour soon, my friend 💪 💙
In case another one come here with my same problem, searching for this EPIC 80s hard rock, glam metal style, and wants to be signed by a real 80s singer! here's the prompt they need to use! Finally fixed my problem:
80s Glam metal, 80s Hard rock, classic 80s rock,
Yup, mixing generes like that can really get you the specific shade you're looking for.
Using parent genres can help nudge things too if you're in a subgenere.
are you a free user?
hi, I haven't used up all my credits but the remain disappears when the expiration date come, is it the rule?
@untold briar @alpine venture Saw this post regarding how MidJourney tokenization weights can shift depending on what other prompts they're used in tandem with. https://x.com/ciguleva/status/1692342449962897469
Further confirms my reticence to provide "concrete lists of prompts that work" because I have NO idea how they'll be used outside of my immediate recommendation.
for sure, also even without weights, how the model parses stuff isn't just a straight "here's the direction, go!"
it generalizes when trained well, so too, the way it puts words together, for example, is going to have some internal meaning ie you can combine some tokens that mean one thing and get something totally different bc of how it percieves that "grammar" effectively
although, it's still useful info imo
we have to use simple heuristics to make good music
they're over simplification, by I ain't no nerd
jk I am
but what I mean is, on our end we have to balance time vs return; my basic heuristic is "good prompts is good"
if it worked somewhere, it may work somewhere else, or maybe combined with another good one, or with minor changes etc. So finding small things that have an impact help compress more into that style box
like, it's quicker to use Mbubeoid than Mbube vocaloid
and it works in a way I like many times
stuff like that
suno legit just made me a 35 second and a 12 second song. tf. what a waste of credits!
Free users get 5 generations (so 10 songs, since each generation makes 2 songs) a day. You'll be able to make more songs when the credits refresh (which I believe happens at 12:00 am UTC)
Ok
I've found though, we can put a prompt, let's say, reggae, at the very end of 12 other prompts and the song totally turns reggae. So I've started having my doubts of weighted prompts. Overall yes. And perhaps it "bumps" any new or changes in prompts as a priority style for one generation, then reweighs the prompts again. Lot of synergistic things going on. I swear it takes ideas from my other songs and blends into new song sometimes. I find myself saying, "Hey that's a line from that death metal song I wrote yesterday. "
-
was this an extension or a "new song"?
-
Suno's output is most effected by what you use as input, so I'd be curious what you were using as style prompts, or if it was an extension, where you were extending from as well (end of the song? From a fadeout? Extending from silence?)
I know everyone has different interpretations, but even when Suno doesn't do what I expect it to, I can still try to learn something valuable from thinking about why it did what it did, opposed to what I could have expected
for sure, it really depends, and it part of why I like to break them down into smaller bits. I find the combinations generalize better bc the model is unsure, or put differently, you've put it in a place that it can't have overfit on easily bc nobody has written "poly wat pulling" before.
in their original data
Hello suno support,
this question may have already been answered somewhere.
It's about instrumental versions. When I generate stems I get the same version as before (instrumental). And the separate vocals version makes no sense at all because it doesn't exist. I would like the playback (instrumental) and the separate instrument, like with the vocal versions.
Can this be implemented please? Best regards, Chris
there are ton of tools, free, to separate instrumental/vocals from the songs
At this time, the two stems are instrumental and vocals (if any to begin with) . Other sites or software can be used to split after you download. That's where it is at this time. In just "appealing to the masses" the stem splitting option may be minimalized for the moment to at least give someone a karaoke track, which can be helpful, but nothing more at the moment.
Thanks for the answer. I thought it would be easier with suno, since a solo instrument is like a vocal track. The idea was to create karaoke versions for music lessons, without the instrument track.
hey guys, anybody knows if there is a way to force suno repeat last part it did - twice ? by any tag [] or something like this. Just that it repeats exactly the same twice ?
any news about the "negative prompt"?
I don't know of a way. Is this just an instrumental, I assume you're talking about.
nope^^ but nevermind, finally with [REPEAT] and similar tags and some trials managed, that it repeated it more or less close to the same sounding ^^
Any reddit posts or gen-chatter about any features is entirely speculation until you see it listed in #📢┃announcements or #🚀┃updates, or unless someone with a purple name in discord say that it's 100% for sure happening. --EVEN THEN, things are moving so quickly, and things can pivot and change direction so quickly that you might miss some really exciting features if you keep waiting for "negatives", because that might not be in the cards.
Please be on the lookout for the updates and announcements for official news 
Isn’t there a program you were using for this @charred merlin ?
Hi @weak estuary -- I am not a member of the team, and suggestions like this are most effective in #💡┃suggestions-web -- the devs read that thread for ideas 💪
"Stem Separation" in Suno would be better named "Vocal Remover" because it only makes a "simple" Karaoke version -- it is not perfect quality, but I do not believe that Suno v3X series had been specifically trained to do this, it might have just "known how" 👀
The devs mentioned in the #🚀┃updates post that "Stem Separation" is just v1, so very early doors, and the fact they named it "version one" makes me very hopeful that they plan to improve on this -- this is admittedly an "early access feature", and I have no doubt it will live up to the "Suno Quality" standard we expect 😄
In the mean time - I absolutely LOVE a service called Bandlab -- https://www.bandlab.com/splitter
With the free tier you have unlimited access to 4 stem separation (vocals, drums, bass, all others), and can get up to 6 stem separation with their paid tier (I am not a paid member, but others on the server have used it and can give testimonials if they choose 💪 )
You can separate the stems, change the levels, do simple "mixing and mastering" all in one page, or open the tracks in their web-DAW -- no charge 👀
Oh -- and when you're done, you use their AI-powered free mastering service to choose from 4 different styles.
https://www.bandlab.com/mastering
Like... it's annoying how helpful they are. They make me look like chopped liver. 🤬
i actually used the stem splitter for the song i'm mastering right now
It's just too easy... plus being able to directly edit the key signature, bpm and stem levels right from the splitter page, not to mention a lite-DAW for basic looping and quick experimentation.
Like... we get it, BandLab. You rock. Stop shoving it in our faces and try to make the site suck a little bit.
i'll send a dm with how its going
Hey everyone. So my subscription renewed today with about 4500 credits left. i pay for the 10k a month. So simple math should dictate that I should have around 14,500 credits right? why did it stop at 10k and not roll over? seems a bit sussy....
The credits "renew" and don't roll-over between months. This is explained at the bottom of the Subscribe page https://suno.com/account
well dang....thanks for the reply
Hey guys, any sugestion on how to make a real 80s pop song? now that i I can do 80's rock songs with this new v3.5 and they sound great by the way, I want to master another 80's genre, pop. Could someone throw me some tags to achieve this? No matter how many times I put 80s and pop, it always throws me that modern pop style that in my opinion, shouldn't even be considered music. It's anything but music.
Off the top of my head, it's good advice to avoid the word "pop" by itself. Not just for 1980s but any pop genre. Use more specific 80s words. Maybe like "glam rock" "hair metal" "butt rock" "arena rock" "power ballad". Maybe try with dashes instead of spaces for compound words. You can use pop like "synth pop", but even then, use it sparingly.
interesting 🙂 i think powerballad mixed with synth can do a good job
i'll try that one
Why not use the word "pop" it works fine.
thank you! this is helpful. i used pop earlier and got eyebrow raising results lol
its likely too broad
You can use pop, but you need to have a lot of 1980s specific words along with it in the prompt, of generic modern pop will overpower it
Oh, I think I see you meant don't use "pop" as the sole style, use like Pop Punk or Pop Rock. Yeah Pop without any qualifying word can do some weird stuff
hi, I haven't used up all my credits but the remain disappears when the expiration date come, is it the rule?
That is the rule. Credits are reset to 2500 or 10,000 at the start of every month. Credits don't carry over to the next month. If you are a new user and didn't understand this rule, you could try emailing support@suno.com and explain that, no harm in asking
/help
Hi there!
If you’re ready to dive into creating music, head over to Suno.com.
For more detailed information, check out docs-guides-resources.
If you have any more questions, just ask in #general-suno where you can always find our community, moderators and guides to help you with most questions. 😊
I created a song today and I regret not having tested it first on a better sound system. Let me say I love the lyrics, album art and the first music iteration is perfect for the song.
All that said, for the second time in the last two weeks I encountered extension iterations that appear to have painted me in a corner. Specifically, music styles that are based on tags like [Ethereal, Dreamy, or similar] create a tinny echoing resonance in the background that can overpower the singer at times and at other times generate background noise on a nice sound system that is just painful to the ear.
I am using 3.5 exclusively for generation. The below song starts out well enough but about halfway in and especially at the end it’s rather painful. If anyone can offer a solution it would be greatly appreciated.
Thank you!!
I must confess, it does seem to be a great point. We pay for the credits and accumulating them from month to month only seems reasonable. That said, it’s rare have any remaining at the end of any month but your point seems valid. You are paying for a monthly service and maybe carry over would be reasonable.
I was on vacation out of the country for a month this year. Either a carry over or the ability to place a vacation hold on your account seems fair. Though carry over would cover all bases. 😀
are there tags for "NOT overproduced" ; "NOT overmixed" "Vocals not auto-tuned to soul destroying crap" ? just..you know..as examples
I know some people do use extensions with smaller intervals (e.g. every 30-60 sec), to keep the chance of distortions very low, for the song idea they really love and wants to keep it relatively high quality.
Expensive credit wise but the quality speaks for it (assuming you had a great quality generated at the start and kept distortions low when extending)
my average iterations on all songs ranges from 6-13 extensions with song lengths from 5-14 mins. I usually try to limit my extensions to no more than 8 lines or two verses. But there is still something in the AI generating excessive noise. The song I referenced above generated atleast 3 or 4 extensions that were so loud from that background noise that the music was actually painful to the ear and inaudible. Just wondering if others are having similar experiences.
one of the tags I have seen that has good results is Minimal or Minimalistic in the style. I think it also helps to reduce the number of tags as the AI probably wrestles with your intent.
In my experience, tags like [Orchestral, Dubstep, NDAI, EDM, etc…] will almost definitely give you what you do not want.
Not in that sense, no. But you can indicate the kind of vocals you want more specifically using phrasing like "solo vocalist, clean vocal" etc... be creative. As with everything Stable Diffusion related, trial and error and luck of the draw is a huge factor.
I don't know of a general purpose word you can add to any prompt to reliably improve vocals. It's most like a bag of tricks. But if I had to pick "one weird trick" it might just be adding "vocals, vocal" by themselves. "vocalist" I can't remember if I tried but it make sense. "clean" I have had bad luck with. I know a lot of users who swear by it, so maybe it does help in some situations, but my personal experience is "clean" is worse. "clear" isn't super obviously helping, but could be
👍
I am determined to use all my 10,000 credits for this month before reset tonight. 850 credits to go 😀
good goolly i wish i had that issue
@dim glacier I want to suggest a change in the export feature (so the filename or metadata includes the identifier of the song) where is the best place to suggest it so the team sees it and considers it?
If you do not know what to do for some of it, I highly recommend generating instrumental pieces of various (sub)genres.
This way, you only need to change the genres and you can use whatever sounds great in a future project.
I personally do not like using the instrumental switch as it usually ends with a short piece or have hallucinations.
Instead, I use instrumental metatags for more control. As an example of my very generic use of broad instrumental tags (opposed to defining specific instruments), you can see it in the lyrics of https://suno.com/song/990073b4-8887-4441-8241-d5374641d1c5
You can suggest that at #💡┃suggestions-web
should I mention someone? won't it be lost in all the stream of messages? 😅
Nah the team look at suggestions there, along with bug reports in the other channels.
yay vomiting stopped
@untold briar https://suno.com/song/323c2ae8-f1eb-4003-9cd0-cec50afa62bd
this is v poetic for your usual stuff, is it from something, or process?
vocal, vocal, vocal, vocal song. Listen and make your own with Suno.
either way, its nice, opens very interesting and progresses into fallout 3
which happens to be my fav video game soundtrack
It's the "let Suno hallucinate, extend the song based on what you think you hear, until it sort of makes sense" technique. and then that technique is incorporated into what the song is about, of course. i do tend toward meta
With just a few direct lyric prompts where a push was needed
👍🏻
Damn still can't figure out how to make vocals cut out its own frequency window
wooh
I guess I can do it "in post" (VFX joke)
All of the songs I publish outside of Suno have a lot of extensions, but those are the songs I put a lot of attention and care into.
has anyone have prompt for meditation music? I want to create it without lyrics for meditation, zen but Suno cannot create it's correct. pleas help
/help
Hi there!
If you’re ready to dive into creating music, head over to Suno.com.
For more detailed information, check out docs-guides-resources.
If you have any more questions, just ask in #general-suno where you can always find our community, moderators and guides to help you with most questions. 😊
While you're able to leave the "instrumental" toggle turned on, Suno will occasionally "hallucinate" lyrics -- from my observations, one of the largest contributing factors to the frequency of these hallucinations is the character count used -- the closer you get to maxing out the character limit in the 'lyrics' or 'style of music' fields, the more prevalent these weirdnesses become.
https://tenor.com/view/mr-bean-mrbean-bean-mr-bean-holiday-mr-bean-holiday-movie-gif-11625313334009911234
Im Looking forward to SUNO 's Official API. There are some suspicious Unofficial api tools in Github..
We neeeeeeed official API!!!!!

We know. We know.
Unfortunately it's not something that we're about to provide any sort of actionable advice in tech-chat about, BUT https://discord.com/channels/1069381916492562582/1253766525458841610 is a known-suggestion post so that you know the devs are aware of the issue at least, and if you want to be proactive about the idea, the devs occasionally poke their heads into #💡┃suggestions-web for ideas -- let them know WHY you want an API and tell them what kinds of cool things you would want to see done with it, because it might be that they're ABLE to provide an API, but no one has explained why they should partner with them in that way, or loosen their control of how Suno is being used while it's still in a VERY early doors period
What prompts do you use to achieve this kind of music? When I try to generate it myself it becomes too epic or too dramatic with too much instrumentary. (battlemusic isn't generated)
why is it taking 15-20mins to gen 1 2min song?
Just "meditation, meditation, meditation, meditation" works, with blank lyrics. But it's maybe too boring. You could add some related words, but it's still maybe a little tooooo meditative: https://suno.com/song/d7af0dd7-8f3c-489c-a59c-a0ac105bafcb
can not run
ah so that might be why I just got Application error: a client-side exception has occurred (see the browser console for more information).
It says right in the box. Elevated usage. Too many people generating at the same time, peak traffic of the day probably.
Oh ok
maybe this has been noticed before, but I imported a Suno track I made into Logic Pro X and did Smart Tempo Analysis and was surprised to find that the BPM is changing so much (this is electronic music, I think I even put the tempo in the prompt) - first, small fluctuations (maybe just Logic being sloppy but I don't think so), then more dramatic changes
Would be awesome to be able to choose the tempo. As one track is ultra fast, one track is slow, i dont have control over it, prompt is ignored
also, to choose the "temperature", like from a chillout, tempting, calm, sexy girl voice, to full of rage, metal-like, ROOOAR of some hardcore guy 🙂
both can be implemented in form of simple sliders, range selector 🙂
While tech-chat is more figuring out how users can help eachother with the technical side of things, #💡┃suggestions-web is a great place to share ideas for what you'd like to see in Suno!
Since the devs pop their heads into #💡┃suggestions-web for ideas, maybe you would be willing to share your ideas there as well? 🙂 💪
there is so much spam, i left my 2 proposiions there, but some kid then commented my sugestion with 200 comments, and after all when i asked if staff is reveiving sugestion channel, he said probably not because i left wall of text there 🤦♂️
but ima ping our discussion from here, to there, thanks for sugestion mate 🙂
Yeah, it can be a little chaotic in most of the channels from time to time 😅
🙏 You're welcome 😄 100% wasn't meant to tell you that you weren't allowed to talk about that here, but in my (probably incorrect) viewpoint, "tech-chat" is like the laboratory where us weirdos talk about what happens if you pour "math-rock" over "indie", etc, while the #💡┃suggestions-web, #📱┃suggestions-ios, #🚧┃suggestions-layout, #🪲┃bugs-web and #📱┃bugs-ios are more like those "Comment Boxes" they put up in Mom & Pop restaurants for you to give feedback.
The devs are super busy, so they don't always have time to pop their head into the science lab, especially knowing that we'll ask them to stay and bug the hell out of them with questions they either 1) aren't allowed to answer, or 2) are still figuring out themselves 😂 So it's not a means of stifling conversation, but just making sure that your well-thought ideas are also placed in the "official comment box", so that way if they go ahead with the idea, you can "prove" it was your ideas as well 😉 💪
kinda new to the server, where would be best to post my observation about the BPM fluctuation?I'm not even making any feature requests I was just trying to share some interesting discovery about the BPM
You're in the right place, my friend ❤️
The moderators (light blue names) are likely better to give explicit instructions as to what the different channels are intended for, since the server is still a work in progress, but from a fellow user's perspective;
- #💬┃general-chat for the most conversation and personal song promotions,
- #🗣️┃chill-chat for whatever (but it may still need to be AI related somehow?),
- and 🎯 here in tech-chat for anything that we can actually sink our teeth into and try for ourselves (like, even if there's a weird bug, it's really helpful to have clear instructions for reproducing on our end so that we can see HOW widespread of an issue it is, or if "it's not a bug, it's a feature", so the bug reporting channels aren't the ONLY place you're allowed to post bug related stuff, but if it's just simply to say "this weird thing happened" without any further attempts to resolve the matter, the bug channels would presumably be a better fit.
just my opinion, ymmv, without the express written consent of major league baseball
cool thanks... yeah i don't know if this is strictly speaking a "bug" just found it surprising that it can't maintain a perfectly steady BPM even within one generation (I'll do some more "experiments")
It’s certainly not unheard of for Suno to start speeding up in the middle of a generation, I tend to get that a lot during chant-based songs, but people who do EDM report it as well.
I haven’t done too much research into what might cause it to happen more often than not, but it’s safe enough to assume that it’s related to what we use as input; and if we really don’t like the way, it sounds, we can get it back on track with an extension placed before the speed up
yeah i imported/analyzed a few more songs with Logic, and I think some of the weirdness was Logic's fault. the big climb in BPM at the end of the image I shared was probably from an extension job or something, it makes sense to me that maybe it can't keep it steady over different iterations. or it just chose to do that for stylistic reasons, but yeah overall many tracks are pretty stable tempo-wise I think
Most are stable, but it's trying to figure out WHY it speeds out that's the real pickle -- while I still don't feel comfortable brushing it off as "just something that happens", and while I still don't have any conclusive proof, I've noticed it happens more frequently when people specifically try to prompt for specific BPMs, whereas there's tools even built into bandlab's splitter that can control the BPM - #📚┃suno-school message
I have a feeling when people say "bpm 120" or something similar in their prompt, Suno ignore the specificity of the request and might just focus on "BPM" as a keyword that seems to indicate some sort of tempo -- and increasing the tempo over time might be a somewhat reliable expression of those hallucinations.
But up to this point, there's more work put into figuring out Suno's successes, but there hasn't been as much productive research into replicating quirks in order to really understand the hallucinations -- beyond simply noting that there's certain types of inputs that can lead to "hallucinations in general", ie. filling the character count, or overloading metatags with more than 2~3 words.
I have a question concerning voice uploads and their impact on voice generation. I have been trying for a long time to find a voice that closely resembles my own but have consistently been unable to achieve a satisfactory match.
Recently, with the introduction of the upload feature, I uploaded several voice samples, which I used to generate extended songs and numerous tracks. However, today, I created a song without using any uploads and was surprised to find that the generated voice was remarkably similar to the one produced from my uploads, even down to the distinct microphone quality.
This raises the question: Does the system train on the files that users upload?
Tagging some friends here for notifications.
@tulip turret @amber vigil
You had a generation that sounded very close to your own voice?
How can I find a song I disliked from my creations
I mistakenly clicked dislike. I checked Help section, but there is nothing about it.
Simply go into the library 🙂 Disliked tracks are only hidden in the Create section
If you are playing the song currently or have accessed its page, you should be able to remove the dislike using those two sources as well
I would HIGHLY doubt that they're using user-submitted content in ACTIVE RLHF (reinforced learning from human feedback) -- "Microsoft Tay" is a great example of why companies don't really do this.
If there's some curiosity regarding a song you generated, it's absolutely worth discussing, but to keep the conversation grounded without too much speculation, there's a LOT more leading me to believe that user-feedback (...>Report>Bugs>...) aren't even used in live training, and only are used for considerations in training future models.
That said, I haven't looked into the ToS to see what kind of rights you give them by submitting this content, but there's way too much to consider for a single instance of oddness to be a smoking gun for something that extrapolatory.
IMO this is kind of an example of what’s weird about Suno/music-AI, the APIs are much more opaque compared to other generative AI tools, it’s much more “roll the dice until you get something like you like”… is this possibly intentional, like “if we expose the model too much then we’ll really be in trouble”(??) idk I need to study the architecture a bit more
“Prompt engineering” music feels much more like guesswork
Ask 10 different twice and get 30 answers 😜
Personally, I think any disconnect between "what it does" and "what we think it should do" is more a failing on our parts to reapproach it from a collaborative level, and comes from a forgiveable viewpoint of "this is a tool made by humans, so it should work FOR humans".
It's a bit "flowery" but I have an analogy I've been sharing in DMs, 1 sec
When getting familiarized with creating a genre or style with Suno, I feel it is like walking through an unfamiliar room in the dark -- if you try to walk too fast with boorish confidence, you're going to smash your toe into something and call Suno a sonofabitch 😂
But if you start slowly, get a feeling for things, eventually you can walk through similarly arranged rooms at a safe pace, and not feel entirely overwhelmed. Eventually your eyes may even adjust to the dark.
This doesn't mean that you "understand all of suno", because when you try a different style of song (going from rock to rap), you're entering a whole new darkened room, and if you try to walk through it the same way you did for a rock song, you're going to smash your leg on the coffee table, and get upset because the coffee table wasn't there in the rock-room 🤬
My work with "indie math-rock" has been more akin to daft punk / julian casablancas / delicate steve, with unique and unexpected song structures, so while I haven't "familiarized myself" with "dark room" to make something sound like the band Polyphia (technical math-rock / metal), I might be able to move at a reasonable pace because it sort of overlaps with what I got with "indie math-rock", even though I wasn't "feeling around in the dark" to understand a specific Polyphia-style sound to begin with.
I agree that it's 100% confirmation bias if you include 10x prompts and don't understand any of them in isolation, and while it's STILL a huge amount of experimention to try things in isolation like I suggest, I did a lot of my "experiment sprints" in the last couple days of my credit cycle so it felt like making effective use rather than wasting credits.
ctrl + z
don't use hashtags with a trailing space or woodhouse gets angry
also, don't try to use subscript (-#) or he tells you that your tiny line is too big. 🤦♂️
Failing all of that, reach out to a guide or moderator to repost your message ❤️
Wait what? lol idk what happened I’m just on my phone…
