#📚┃suno-school
1 messages · Page 150 of 1
Its a new model, it's in beta, there will be a lot of improvements that need done.. Some sytles and genre are struggling, others are just magic now.
A lot people mentioned something like this
For example, in my isekai'd Woodhouse song, I used the following in the style prompt:
A very Catchy upbeat J-Pop influenced EDM mixed with Future Bass and end the song off with a crescendo and a final bang
Then I clicked the magic wand icon until it gave a description of what I wanted and it turned out pretty well
https://suno.com/s/cM8s6hE7VWqL0axo
Perhaps this workflow entry can help you? https://discord.com/channels/1069381916492562582/1367549634536542389
I get "New", great, then let the free users figure out that it has issues, don't be charging me for it. lol
You dont have to use it... you still have v4
The main problem with free accounts is the bot abuse issue, one reason why I assume v4+ are still restricted to subscribed users only
To make it mash up and morph one style into another, I'll assume it could either be done by changing prompts in the style box as you add an extension... or...
Use metatags at each interval (i.e. stanza) in the lyrics box on the initial new run, where you want an exchange.
The promotional advert example seems focused on the Simple prompting where you just tell Suno to create a song and "use these transitions" and let Suno write its own song. So it seems up to us to see how well it works in the manually Custom panel.or style prompts and metatags.
I know, but 4 has been changed as well.
Ye will test it
Honestly, I really like having limited access to beta models, we can shape the direction by leaving feedback about things we like, dislike or have ideas for improvement.
I understand that, but I do AI for a living, lol. Testing = work to me.
But every time there is an upgrade, they also downgrade the lower model to attempt to garner sales.
A professional tool this is not, its a tech toy at best lol
Hi all, I've been messing around with v4.5, and I feel like the sax and guitar solos don't really sound like either instrument. The sax on v4 sounded pretty good, like when doing solos. On v4.5, they don't sound organic. They sound artificial. And guitar solos aren't really solos or guitar. They're just a sound, which I know doesn't make sense. Point is, those instruments don't sound like they should lol. Anyone else experience this?
I also use it for professional reasons.. I think my account is at $400 for Apr
Beta = testing Doing AI for a living off of only Free services is quite a limiting career. You'll constantly lose momentum while others leave us in the dust. Gotta spend money to make money.
I write songs for a living and know i have to pay to play.
I fully disagree with this.. v4 has had issues since long before 4.5 was out, they didnt nerf it to get more subs, otherwise they would have made v4 available to free users
When 4 came out, it was crap too, but it got its place, I just wish they wouldnt touch the others until its working.
That's basically how it feels. It does smooth transitions from one style to another but you're out of luck if what you actually want is a mix of simultaneous styles in tension
Hey guys, just want to verify if he following could happen to me? My lawyer friend tells me that if Suno gets fined or loses the lawsuit, I might also get punished? He told me that the reason Suno is requiring credit cards might be for KYC (preventing money laundering) and for passing litigation charges and fines to customers that violate "ToS"? Like you will be the one to get fined millions if you're the one who generated the copyright violating song? Does anyone know any legalese, and give some assurance that this probability is impossible or unlikely?
Please go here #📝┃user-research-feedback message and leave your feedback!
You can disagree all you want, doesn't make it any less true lol. Just like chatgpt, the free version seems to get "dumber" every time they push out a new model, its done on purpose, Suno is owned by Microsoft, Microsoft's AI is ChatGPT. do the math.
Does anyone know when we'll be able to download songs via the android app? 🙁
That is not true, I had confirmed with the devs themselves. As soon as the models are marked as released, it is frozen and untouched unless a huge issue appears.
Chances are it has something to do with your prompt (most of the case really) or RNG. A rarer chance would be modification to the interpreter.
Regardless, as Ginger mentioned, you can leave your feedback on the form provided
None of here are ( or likely arent) Attorneys or judges, we cant answer that kind of query
And you can believe anything you want; but making right decision in your business is also critical... or if you want to improve it, give more constructive feedbacks rather than speculate ideas about making the older model dumber to sell newer model, you have no evidence just opinions.
Also, MS has an investment in Suno but its not "owned" by MS.
He said something like with the Spotify botstreaming case, where the distributors got fined, and the distributors charged the cards of the suspected botstreamers hundreds of thousands of dollars.
None of us here can offer legal advise
If you didn't do anything illegal (e.g. used copyrighted material in your own output), you have nothing to worry about. The Suno/Udio lawsuit is about use of copyrighted material for transformative outputs. If you uploaded copyrighted material for your song generation, it would be derivative (as it is based on that song) and the latter has been proven in court time-to-time that THAT is against the law.
The way that case will impact you would be in regards to copyright protection, they will not attempt* to sue everyone
(* Personally not sure if RIAA will try. They are the ones that wants easy money by threatening to sue just so most people settle out of court, but generally, you will likely remain unnoticed)
Fair point. It may move some resources to the upgrade and leave whatever is "less" for the Free users, which comparatively is better than 3.5. Clever marketing at minimum.
It's good to know that. Budgeting $10 a month for a paid tier is a great workaround, if you think it will make at least $20 a month extra profit. Smart business choice for an AI career.
That's exactly what I'm worried about. I can't control whether Suno generates copyrighted material or not. Like if I use my own samples or compositions which follows a common progression and Suno uses that instead to generate a copyrighted melody.
To be blunt, I don't see how it's reasonable to build anything on shifting sands and the whole v4 arc showed us we can't really rely on much besides a short window of time where a model is usable before it becomes trash
Just talking to myself: For what its worth i still use 3 and 3.5 and certain to use 4.0 as times goes on, for the "talents" that these versions seem to have over the other, until they drop support or just remove them.
The Suno/Udio lawsuit, at the moment, is about use of copyrighted material for transformative outputs. If you uploaded copyrighted material for your song generation, it would be derivative (as it is based on that song) and the latter has been proven in court time-to-time that THAT is against the law.
As for your samples, I mean, why would you expect to get sued over your own samples? Unless it is a sample from another service and you violated their TOS, I would not worry about that - it is akin to worrying about your router giving you cancer
In August will we be begging to return to today's 4.5?
I will absolutely still use v4 for certain genre, until a new model improves upon them.
You're right. Good points if true. I agree I can't rely on everything here to work as I want or even expect. The resolve has been to increase my options at my cost. A necessary evil, if we want to see it that way. Or a profitable secret - find all the hacks in the unreliable -- and succeed.
I've actually increased my "work" hours and budgeted back some other expenses to make my membership here and elsewhere a priority.
That may not be for everyone. So I get your point. It's hard to work on a tight budget that the need to rely on consistent performance becomes a priority as well as a hindrance.
Thanks! That channel doesn't show up in my list, and it says I don't have permission to post there. I don't have permission to post in the bug channels either.
Here is the direct web link https://sunomusic.typeform.com/to/hcFMxd2L?
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
Just in case you were unaware: As far as I know, all Music AI generators stipulate that whatever input you use for creation of the media, you grant the company and affiliated third parties irrevocable rights to use and modify those inputs at no cost to them.
So if you use your own samples, those companies can use it without paying you. Will they? Very unlikely but that is what you agreed to when using those services.
Top Music Attorney recently put a video about Udio partnering with Audible to make fingerprinting of the AI generated songs, which means if you use your samples in an input, it could be caught up in that fingerprinting. https://youtu.be/8c9hrHkcWCM
Suno also has a partnership with Audible but that is moreso to check for copyrighted materials - at this point anyways.
https://suno.com/blog/suno-partners-with-audible-magic
Miss Krystal knows her sht.. she is a gold mine of info
How do people fully remove styles. Even the displayed ones have the expandable original styles
I don't think that's possible.
what is up when i publish some songs and i leave the page it just erases them and dont show them anywhere but on the personas...they are not on my page at all but they are Public
There appears to be some web loading bugs.. most pewople experiencing this find everything back later.. sometimes minutes, sometimes hours..
Looks like Ginger got you an answer here, apologies for my lazy Saturday
I'm not worried about my own copyright, but I'm worried about getting dragged by Suno into its legal issues and getting punished and fined, just because the sample/beat that I came up with causes Suno to uses copyrighted content to generate a song. Like anyone can be inspired by a beat or melody that they had unconsciously overheard, then Suno uses that to interpolate/predict the arrangement that the copyrighted song uses instead, producing a similar sounding song. Since Suno has my card details and information, they might do the same things that distributors did to their customers who they accused of botstreaming. Anyway, top music attorney seems interesting, I'll take a look.
ohh okai thank you Ginger!
ive learned with A good prompt you can copy just about any artists Voice lol you CAN clone haha hopefully that gets fixed soon to avoid any lawsuits
EXAMPLE: Boyz 2 men, ((MY LYRICS)) https://suno.com/s/bl4AL4hbs0MTZQc1
You do not have to worry about that honestly. It is like getting sued for you personally drinking a beverage from a company after they had lost a court case. Sounds silly and like it will never happen when put it that way, right?
The things that will put you at legal liability, regardless of the result of the lawsuit, would be if you committed a copyright infringement - whether it be a sample or the actual media or lyrics due to an unauthorized use of a copyrighted material.
I can definitely see major label companies one day suing Music AI generators for a list of users and personal details which have generated music based on those violated copyright materials (and they likely won't get it until subpoena to court, only due to the discovery phase). Well, that or the copyright holder reaching out if you actually published said material...
Avoiding much legal or political discussion here: My two cents anyway.
Money talks. So many litigants attempts are at settling and controlling a bigger peice of someone else's pie. That includes the customers already baked into the pie. One hopes to not bite too deeply , so that those tasty customers are still there to continue profiting from.
Sorry, I'm hangry. Its lunchtime.
i still don't get why covers don't remember the BPM of the original song
can someone help me?
BPM and Tempo are still big issues in 4.5
tho i tried to put the exact bpm like: 102 bpm and used the creative boost and it seemed it worked
I had had very little luck getting Suno to respect my BPM or tempo direction.. it overrides what I want if it thinks its outside teh norm fo rthe genre
i’ve got some feedback on 4.5 - love the new model, the 1000 char prompt is sooo welcome and adherence is improved from 4 …. that said looks like my output is lacking a lot of dynamic range, which would potentially point to either something i’m doing wrong or the model outputs at a lower sampling rate then over sampled to 48KHz …. either way i’d love your thoughts on that, right now my music feels muffled awwwwwww
As far as I can say for now, I notice faded vocals, odd instrumentation, muffled sounds and the sort on many Remastering of 3.5s or 4.0s to 4.5. As for originals, not so much or brilliantly great. Time will tell, but always having fun.
for me it’s both — the remasters from 3.5 are actually in better shape ahah
Please leave your feedback here! It helps us all! https://sunomusic.typeform.com/to/hcFMxd2L
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
Anyone had any disappearing gens? It seems if you cover a song without a persona or style it'll generate but then disappear 🤷♂️ Had some really nice generations just vanish 😭
I checked trash and my library And couldn't find them, then I got reimbursed the credits
seems a web issue, give it some time they should just reappear.
For such a scenario, I would hope that there was a contract or paperwork trail showing that you paid the person for what their work detailed, that is what would make short work of that case.
Always have some sort of paperwork or contract when working with other people on a project. CYA
If let's say that such a thing was not in place and there is nothing to link it to the other person (no bank statements, no saved online docs, etc.), then you would be on the hook since you have no evidence aside from your words that the other person was culpable. It would be like saying "It was not me, it was my [relative]" - it could be true but definitely sounds like shifting the blame.
After all, courts are not really about finding who is really guilty or not, it is about who can have sufficient evidence to prove/frame who is not/guilty.
For that later example, you - the author - will need to sue your friend in another court case to recoup your losses; it will likely not impact the copyrighter's suit against yourself as you are the named author.
(And the publishing company will be sued as well for not doing their due diligence too)
Ill add - and only regarding Remastering. As I've done many live performance outputs, most of my 4. Remastered songs containing spoken parts are getting garbled, or not sounding as authentic, a crack instead of a clap, a rush of noise instead of laughter. It has already seemwd more consistent issues in 4.5 remaster than in 4.0, but typically all my live or spoken outputs have needed to remain 3.5 which works fine. I'd just like to have options in house than to download and DAW.
Try if it sounds like that when played on your phone as well, on my system all the generations also sound bad in some way but some sound good when played through my phone with headphones which also points to reduced dynamic range to me
I have a question, can you make a "persona" with a "cover" but with the fact that the original track that the cover was made is inspired with an upload audio ?
From an attorney perspective, it's evaluated for how difficult can it be made to prove. How grey can we make the area so that a settlement far ourreaches the cost to continue. For $20,000 I agree to settle by adding the other authors name to my forward and 10% of my profits. 😀😉
Mmmm... "inspired"? I think I'll want to skip this question at this time. ( still at lunch) 🙂 anyone else?
I don't think they will as I got the credits back. It never come up with any errors and I generated more straight after with persona and they've remained. It's only the ones without a persona or genre style that have gone
Did it generate fully to a timemarked artwork? Could you listen at least once to it?
Yeah I was listening on repeat, I had the option to remaster them, so I did but they never generated. which prompted me to refresh the page and then that's when I noticed they were gone
So it was after the attempt to Regenerate an output that they disappeared?
Did it generate the Remaster fully to a timemarked artwork? Could you listen at least once to the Remastered output?
Well after I refreshed the page they disappeared.
No they never generated at all.
Oh wait..you're saying the original songs you had been listening too ALSO disappeared AND the Remaster never generated?
basically yeah
The originals ones you listened to on repeat? They basically disappeared - just after refreshing.
yes exactly
Just clarifying what "they" are. It's odd an attempt to use a tool would also wipe out the original output, so this sounds important if true.
I had just normal instrumentals disappear by night yesterday, tag was "OYOJEE" for testing purpose haha
Maybe there are banned tags
xd
I think it was something to do with it having no style or persona
Oyo had one song he wanted to summit to sunovision that vanished as well
If so I didn't make it on the list as Gallis works 😂
KEEJOSX works as well, I'm too "nobody" lmao
Im gutted because I know I won't be able to re gen it 🥲
That's no different than hitting the "permanently remove" button, if we don't find original songs ( say from last week or last year) in our trash bin because an updated tool wipes them out.
Weirdest is that Oyo song disappeared after he uploaded an animated cover and then the upload bugged out the gen
Few minute later, vanished
Ohh damn, that sucks. Wonder what's going on
I'm not generating anything for a week I think ><
Might be talking about two different things. I'm hearing of new outputs within the past 2 days being generated, played and then later not found.
This scenario with Gallis is as if randomly taking an old output, by then attempting to use cover or a personas, literally wipes out the original song.
Yea that's even worse
I don't think it was the remaster that done it, as there were others that I didn't remaster that also disappeared. That also had no style or persona
But I mean something is linking all these vanishes, the root of the problem is probably the same
Right. I didn't mention Remaster, but I got it. Something strange, no matter. Your details though are very helpful. Thanks.
I was able to create, this morning, a song locking up , that some are having a bit of time with, but heck if I think I can force one to disappear. 😲
Hey guys, I know it might sound like a weird question, but have you noticed that Uno deliberately screws up one or two letters in words, here and there in a different parts of the song?
Every song that I generate I have to fix one or two words in the song. And it takes multiple attempts.
Then I go an apply the edit fix and it fixes it.
It happens practically with every song that I make.
It seems to me that Suno does it deliberately so I can spend more credits on applying edits to the song.
I cannot think about anything else at this point because I contacted Suno multiple times but they never got back to me.
If anyone ever encountered this issue could you please let me know if there is any way to fix it?
Any opinion or advice would help.
Thanks a lot guys!
Do covers use more than 2 minutes of context with 4.5, or are we still stuck at 2 minutes?
I got up to 3 mins on a few, so it definitely creates longer covers. not sure what the limit is though
upload with more minutes can be possible?
The limit is 8 min and to bad u cant extend because its not working
The longest i have made is 7 min
v4.5 does Dark Cabaret so good https://suno.com/s/PPTrdJ0txia7dgea
Ja also good for opera and soon hehe
If u like music with a bit of atmosphere its quite good at it i have made a few with v4.5 and they sound is awesome
Listen to this v4.5 works very well for this
Try to hear the 1 i just postet in here and let me know what u think
Is it possible to purchase Pro codes that I can give away as prizes for a show/contest?
Is anyone else having trouble creating "Personas"? Even though I'm already subscribed, it keeps telling me to subscribe to create them. I've sent an email and also reached out on Twitter (X), but haven't gotten a reply yet.
Dunno havemr made any after v4.5 but 1 thing i do know is that extend dont work well i cant extend my songs soo
I've had this happen and a hard refresh fixes it. If you don't know how to do a hard refresh just clear your cache.
I'm using the new 4.5, everything today sounds muffled. I tried several devices. Even asked a friend for help. I bought extra credits because I just thought the version coming out where not to my liking so 2,000 credits later. I believe it's the app. I tried the website too.
What do you mean with muffled?
maybe this is browser cache related, have you tried it in private/incognito window?
For kicks and giggles, also go to Edit UI and click the playback button there to play the same songs. It was oddly interesting when comparing. For many I played the Edit UI playback sounded better than the normal play area.
you are probably trying to make a persona from a upload with voice. that is disabled... BUT it shows the wrong information banner... it should show "You cant make personas from uploads with voice in it"
Would anybody have a theory as to why every single, totally instrumental audio clip I try to upload gets flagged as having vocals?
I've even tried to upload random noises and get the same message.
sometimes it does that... tip: stem split in suno and reupload the instrumental stem.
I'll give that a shot. Thank you.
if a thing knows whats vocals for suno... its suno 😄
Apparently it still doesn't know what a theremin is! 😂
The new update has gotten it closer but it still sounds more like an electric string instrument
i would say thats a conspiracy ... https://suno.com/s/hO9ZuVkqXyAD0MeR
Like some kind of cyberpunk fiddle
I would say you should change your name to Alex Jones. 😜
How can I permanently delete songs from the recycle bin using the app itself, not just through the website?
No idea. The app seems to lack many of the functions that the browser does has. (Trying to multitask two entirely different thought processes at the moment)
some things just work on the website. usually features that are missing in the apps are not completely done on the website. they get added later on if they work properly.
Emphasis on "IF"
Turns out my phone's speech to text doesn't work so well after having facial reconstruction surgery. Somebody needs to train the speech recognition to recognize Daffy duck
Hi ! Does someone has tips to generate not a full track, but just an intro or an interlude ?
Gen the track then Crop it
Why am I getting "content guideline" warning on public domain work... O Holy Night in this case.
Is it possible to remove vocals from songs? Everytime I try to create an instrumental cover of a song with vocals, suno adds vocals that are gibberish
Use Stem feature to split the instrumental stem, then cover that instrumental track instead?
I'm not seeing Edit UI
ON the song... 3 dots menu > Song Editor
The top option?
It won't click. I'll try on the laptop later
Yes. Best on Desktop. Many features are limited on mobile app.
I try to use the mobile desktop because my laptop is annoying lol , so slow. It's not old either lol 😆
I hear you, but yea, editing is best on desktop because of screen estate.
Thank you so much
Can we get instrumental from already created music? They were created with basic account if that makes a difference
Hello. I need some help. Paying for suno but i cannot download my songs Ive written ages ago yet have only made into real songs recently. How do i get my songs
I am on the app on Android phone
What does this arrangement mean... [Intro – Deep Male Narrator (Whispered Lines) Only, No Female Vocals, No Female Voices, No Female Solos]... it means that we don't want female voices, so why does Suno insist on putting female voices?
You can download through the website and I think on the iOS app, the Android app does not have a download option yet
You mean of a song already published and copyrighted? We cannot help you with that as it is illegal (copyright owners can sue up to $150,000 USD per infringement in US Court, and there are international copyright treaties to get you if you do not live in USA)
No my own lyrics
Android app does not have this feature yet. Just go to the website
But thank you. Its my first time on the website and i accidentally threw a song into work area after it said create movie. Now its gone
XD
Mind you, as far as I know, the vocal labels are more implicitly learned in regards to gendered vocals rather than explicit supervised training, which would explain the past ~60% successful rate for duets I had (well, successful in getting both genders, though half the time it swapped the vocal lines...).
You can do something like this:[Verse 2][Boy] I saw my gal in the kitchen ([Girl] mmhmm) So I told her to make me a sandwich ([Girl] Na haa)Text in parenthesis are used for backup vocals and usually applying the metatag within it worked for me in the past, so long as it was simple.
In addition, the AI may not understand what "No" means, especially in the style prompt where it will assume that it is a style keyword.
If you want to improve your chances, skew the term weight to the gender you want by repeating it three times with commas in-between, like "male,male,male" or "male, man, boy", etc.
Does not guarantee that vocal but nudges it a bit more
Thank you so much
If you want to know more about the weights, I wrote this up during the 3.x days #💬┃general-chat message
Click Menu from a banner or list , not the full page. Choose edit, then crop to get to the page
Note that you will need to be in Desktop Mode if you are on mobile. The Edit UI is not designed for mobile view
Whos going to sue, Suno?
Yes. Desktop mode. VrKing is right. Still, use the list output menu not the menu on the full page artwork.
The people who owns the copyright, e.g. the recording companies like Sony, Universal, and so on
For example https://www.billboard.com/business/legal/tiktok-rapper-pay-sony-800k-illegal-music-sample-judge-1235642526/
I think you misunderstood. I created some music with the basic plan. I'd upgrade my plan to get instrumental out of them
I ain't talking about stealing other people's work
Ah, I did mentioned other people copyrighted material in that post, hence the confusion
But, yes, Suno can sue you as they own the ownership and did not grant it to you. Will they? Very unlikely
Probs not but I'm curious about the possibility. Can we do it or not
Some of the generated beats are phenomenal
Ok im getting a hang of it slowly. Is it possible to shorten my songs. /Trim
Though if you upgrade through that popup when you click the download option from that song, the devs did confirm with me that specific song will have their status changed to commercially licensed.
For getting additional tracks made during the free plan into the commercial license, you will need to contact support. They used to be good about it if it was your first time subscribing, unsure the state these days
If anyone knows if i can trim my songs, kindly let me know where and how 😅
I think you can with a subscription
I have that but i cant find it
Prem subscription not basic
Im currently paying
On paid subs and desktop mode click menu/edit/croo to take you to the Edit UI . Tools are tabbed at the top.
Thank you that is useful!
Crop will save only the part you Crop. Remove then takes out what you choose and saves the rest as a new spliced output.
Has anyone noticed that the edit tool seem almost entirely useless? I have tried at least 50 times to replace a section and 100% of the time it run well past the point it is supposed to replace. It seems like it has no ability at all to generate a section that is the same length as the one it is suppose to replace.
Thank you so much for explaining this
I haven't noticed yet. 🙂 Could be a bit difficult to use, for sure. Not 100% of users like everything offered or have 100% success.
I have got the crop part to work and I think I have got the remove to work at least once, but the primary editing I need is the replace section, which unfortunately I have not been able to get to work even once since the 3.5 model was the new one.
That's been my experience, that it's not reliable in a couple of clicks. Sometimes I may need to rewrite parts or, re-extend if needed. Not optimal for sure. So yea I've noticed it's not efficient, but occassionally effective.
So, i get samples that are the same length most often, but find they don't splice in at the perfect second. I use remove , very carefully and I'm done.
So, almost useless, yes.
Sorry there are issues with yours. I do, as a check, is focus on the time marks, they can change and must be manually entered left (endpoint ) to right (start point).
There was a week or so an issue as well with the "Keep same time" toggle/tick box. I don't know if it was working but certainly glitchy. I'll go check now.
It seems to be clickable at least. Do you click this to set? Sorry, I don't know your experience.
Yeah, I have tried it with the keep duration check and unchecked, but is doesn't seem to matter at all. I have been using it in chrome. Maybe I should try a different browser and see if the browser makes any difference.
And every one is exceedingly longer than the duration 100% of the time? Maybe it's not fixed. Are these before 4.5 was released.
OK. For a test to see if its running again, I'll do one now just to test.
I haven't got even one that came out the right length in either 4.0 or 4.5. It seems like it used to work back in 3.5. Usually they are at least 1 minute but often 2 minutes long when I am trying to replace a section around 30 seconds. It also doesn't seem to matter if the lyrics are replaced. If I am replacing a line in a verse, it does the replacement but then it just keeps regenerating past that point right through the chorus and often into the next verse.
Three duration samples were the she same.. 1 was 1:24 .. I replaced with the words THIS is a test, and it splice correctly. So as a control, mine worked OK. It may be other factors involved. https://suno.com/s/JF8bUahWuTEjis73
I just grabbed one, which happened to be a 3.5 song. But the Edit UI does seem to be functioning , if at least once. Not sure without testing 4.0 and 4.5 if the Edit tools react any differently.
can someone help me with this. No matter the device, I keep getting this
Just want to clarify, that the Inpainting/Replace feature for V4.5 is not functional yet.
I will keep trying it. Again, I might try using a different browser to see if that has any effect. Another thing I have noticed is that when I use a Persona that I made in 4.5 and try to create a new song, the sound quality overall seems substantially lower than when I generate the same prompt and lyrics without the Persona. It seems much more compressed and the vocal seems to somewhat further back in the overall sound. I will test it more later, but just on early uses in 4.5 with the Personas, I am getting low quality sound.
I'm trying to keep mind of the issues (as well as try out some of the cool stuff) but looking for patterns to form to find some correlations, such as whether it's a persona made from 4.5 song or just the use of personas overall, or even the use of anything such as personas AND covers.
So for now, write down some issues that tend to repeat and as you've done, report as a bug and let them look for comments regarding similar issues.
I'm giving it a couple of weeks, trying to have fun writing but also trying to repeat users issues so I get a better idea.
Ill later try your 4.5 personas issue and see what happens. Thanks for any updates.
I can say one thing.. i've used a lot of credits on this V4.5 and the audio, it's not even close to where it needs to be sadly. Wasting a lot of credits or I feel like I am
It's been a coin toss sometimes. Mostly heads, but sometimes tails. Occassionally just lands on its edge and spins. It's good to release early to have a bigger range of users "testing" per se , at our expense, sorry for your frustration from one member to another.
Hopefully they work out all issues and make it the best one yet!
Hi. I don't know if this is the right place to ask.. is there a way of seeing all those who liked a song when the alert contains more than three, such as "and 13 more"?
Still no timestamp usage correct, I mean I love 4.5 so far
Is it just me or are Personas broken right now? I've been subscribed since last September or Octorber. However 4.5 personas are saying I need to subscribe to create one... I just tried upgrading my sub to the $30 per month level and they still won't let me create one.
Are you trying to make a Persona on an uploaded audio with vocals detected?
In addition to not being allowed to be made public, they are prevented from having personas made (due to likelihood of impersonation)
It was a Rap Country Fusion I sang my self created by ChatGPT. But what I was trying to do the Persona on was a Cover I did of that uploaded track.
So does anyone have any advice on pacing? Cause I feel like the new version wants to rush the songs whereas version 4.0 at least paced itself decent. Sharing this as an example I'm working on original takes of old songs.
https://suno.com/s/9ss8rutQjNFCPDQW
The original versions had good pacing. Just feels like now, everything is rushing.
anyone else having issues with generated tracks playing and downloading
If you're using a browser, you can either right click or long press to open a sub menu that allows you to open the new song editing extension.
Then on the left of that new interface you can click the traffic light to select "match lyrics to audio."
So far I've had pretty good luck with it although it would be nice to find a way to trim portions of the song instead of only being able to cut clips.
Too many ways to extend the track but no way too shorten it as far as I can tell. 🤷
Good thing AudioLab is only a $1.49 a month
Suno failed to recognize "blown bottle" as an instrument
blown bottle and musical jug are related instruments
Tuning is done by adding water to the instrument
It also thinks a uilleann pipes is a tin whistle 😛
Just have to accept that rare intruments are not going to get represented
The blown bottle isn't that rare - unlike the jug, it's part of MIDI
You dont hear it in mainstream music, which makes it rare
also, it fails to understand what a bass-piano lead means (a piano played on the first two octaves as the roaring bass instead of guitar)
bullcrap
It also has trouble with steel guitars as well as steel drums.
Makes me wonder how it interprets great highland pipes.
it has no trouble with handpans
Or pan whistles for that matter 👉😉
I accidentally created a yanni song earlier and it made me want to cut my ears off and throw my speaker through my lobby window
I started having Avatar PTSD
Handpans are those UFO drums.
Has anybody tried an ocarina prompt?
stainless steel or nitrided steel
Now I need to go see if I can make it play a kazoo and a slide whistle
Clay
Oh, you meant the drums, didn't you? 410 stainless
Got to make that rhythm section sound like I'm destroying my kitchen appliances
stainless steel, nitrided steel, or the tongue drum variant - different note thing shapes, different materials, still the same steel UFO.
Have you tried prompting it to do a steel UAP instead? 😜
Nope, but "handpan" did work
I could have some fun with that actually. Especially since I saved the nightmarish acid trip it gave me last weekend during the update.
though the handpan's resonance drowns out the brass
Even if you were to use a sousaphone?
Apparently there's no way to change the order of the layering of the instrumentals yet.
Maybe try using different euphemisms?
May I suggest using something like, "features heavy brass rhythm accompanied by a harmonic handpan melody"
Alas, my credits are all up (free plan)
Ah, that sucks. Well, there's always next month I suppose or you shell out 10 bucks for the top tier.
Cheaper than some of the games I've subscribed to in the past lol
ahem, it's not the amount, it's the georestrictions
And since you own IP rights to any music you create in the top two tiers, you do have a potential to at least make up that overhead.
However, I am paraphrasing and you should definitely read the eula before you monetize anything.
Pro tip: US copyright office allows 20 compositions per album per application.
I've actually been making progress there. Bye primary philosophy in life is FAFO
Sorry speech to text with a speech impediment
I actually am a chef so I try to keep my hands off my phone as much as possible
I'm not ready to publish this one yet, do you mind if I DM you an example?
maybe...
really, a toothpaste colored sax is an obscenity that shows the art generator is bad
What flavor toothpaste?
Is it like the striped Colgate stuff? Or the sparkly crest stuff?
The track is "toothpaste blues", and the cover is a white-red sax
That just hurts my sense of design aesthetic
Sax striped like Colgate toothpaste... grrack.
Yes, that is bad - and that's why the art generator is a big poo
Bet you they stuck the reed in the wrong way too
Like when my younger brother bought a used sax from a garage sale and then asked me if I knew how to assemble it. I play trumpet, and guitar, definitely the wrong person to ask for that lol
My point is, the AI's cover art pipeline is bad
Oh, absolutely! But, after all, that's not really its primary function. Don't even try doing anything with the videos
So, you want to hear a decent cover, or did you decide against free inspiration? 😂
Maybe I'd try to get the VIP sub one way or another, if I could simply connect a different art pipeline (and have more credits). For now, though, I'm sticking to the free plan.
Fair enough. There was a way I could gift you a few of mine just for the helluva it, I would. I'm sitting on just under 2K right now.
*if there was
On MTS Pay, I can generate those vouchers - virtual Visa cards that you have to spend in one place - but those only come in $10, $15 and $50 (no $20 or $30 intermediate)
so...
yeah, way too much hassle going that route. Either way, I've been building an infographical workflow to share with everybody which I believe will greatly help the community
I might wind up having to make a slideshow of the 50 or so screenshots I've been saving
if they did add external art APIs, instrument layer order...
I miss using PowerPoint
Something like that probably won't be till 3 or 4 versions down the road at the bare minimum
So, no Pro for me for 2.5-3 yrs
Affirmative. And that's only if they feed their hamsters well.
It takes a lot of those little effers too power a server
🐹 All
🥣 gone
Are you also having issue with personas for 4.5 that always tries to have the same melody?
Same issue for a few weeks now.
No one really know how to help, some advises to clear cash or cookie, to try other browsers and devices, write to support and just wait
Is anyone able to create a persona that actually sounds like the same artist or band? Want to create an album using a persona. Also wondering why you can't create personas from covers of uploaded audio songs
That's not quite what I mean. The words match the lyrics, it's just they're almost trying to be a whole run on sentence.
Unfortunately, Suno may not be as successful as you would like.
If you want a very good replication to have a consistent AI singer profile, you will want to look into "RVC Models" to make your own vocal profile. Then look into "Vocal swapping" or "Vocal cloning" services which can use the RVC models you generated to replace the vocals in the song.
There should be several web pages and YouTube videos about this. You can search my previous mentions about this as well for an example I linked before
Two old, related links are
#💡┃suggestions-web message
#🗣️┃chill-chat message
Check YouTibe for the latest service reviews for services like elevenlabs, kits.ai, and many others
Got it. I was hoping I could take a "band" I like and it could recreate that band, with essentially the same instruments "band members" so it feels like a real album.
When I say "band" i mean from a suno created song
it's more about fluid creativity and less about constraining or composite part assembly
Ah I see, that sounds a great deal more difficult. I just hope you do not mean a real life band because you can be sued for impersonation and other related crimes.
There are actual laws specific for AI impersonations, such as the E.L.V.I.S. Act in Tennessee, USA
No, I mean from a suno generated song that I want to replicate
Ah gotcha
And even parody and fair use laws can be a little convoluted.
I sub to eleven labs but they don't do vocals as far as I can tell, only spoken
Look how many times people tried to sue weird al
Do you still recommend weights for RVC? Or is there a better one out there
Unfortunately it has been a long time since I played with it and it was a temp. curiosity
There should be some tutorials that you may be able to find online, but chances are you will need to deviate a bit, based on the profile you have
Congratulations! You win a drink from the fire hose!
Is it possible to make a persona from the cover of an uploaded song?
not if it's of voice or voice was detected in it from mid high frequency with voice like articulation
It's been working pretty good for personas created in v4.5 from v4.5 songs, not so good when creating personas from older versioned songs.
Anytime I use custom lyrics they’re completely ignored and completely new lyrics are generated. This has been happening for days, since I started using Suno actually.
I need help logging in
So I’m just forced to use whatever it happens to give me?
are you putting them in the right box
lyrics and not style?
Yes
What prompts are you guys using to stop distortion and clipping with v4.5? I'm getting solid tracks but 4.5 is making the tracks too "loud" and ruins the track
Use the "Custom Mode"
peak, compression, distortion, clip, limit, all in the negative, sometimes adding reverb or for other kinds of poor quality, like a sense the music is hollow feeling, i'd negative prompt wide, wide stereo, if the music is lacking high end, lowpass, if it's lacking bass highpass
I’ve been doing that. But custom lyrics work when I generate completely new. The problem was when I used “reuse prompt” (even though I was still using custom lyrics)
Back up a bit: First, who's output are you reusing when you click "reuse prompts", your own or someone else's?
Hey how come moderation won't let me make a song about this "A horror eerie rock song about 2 players playing zombies on xbox Surviving with teamwork, skill and dumb luck" of course I used our player names but that didn't work either. What's wrong?
TL;DR: ChatGPT, the censor, didn't like something generated in the lyrics. Might be zombies but likely references to skulls or bones. Maybe (attempted) murder
😮💨
is there any easy/fast way to access all the audio clips you've uploaded?
i might've uploaded 100 clips in my history, in different workspaces etc, but i can't see any easy way to find what i've uploaded. am using desktop version, not the app
If you do not have that many workspaces created, you can use the Filter and check off Uploads.
Unfortunately, I am not aware of a similar method through the Library
oh wow, thanks, that's exactly what i needed because most of my uploads were in generic "My Workspace" 🙂 thanks again
You're welcome, glad that worked out
My lyrics despite trying a lot of different methods keep getting cut off in the Lyric Video and I think I know the issue.
For whatever reason, there's a hard symbol/character limit for an on screen line and it's jumbling things around to make that limit per line a hard rule.
I've not had this issue with other songs with even longer lines when downloading their videos, only this one.
It also won't show the first few lyrics.
Here's what I wrote, it's PG. Maybe It has something to do with how I wrote it.
`[Intro]
[Loud Echo]
I know you might be scared,
Let me cover your eyes with my hands,
My Precious one!
[Verse]
[Overlapping Echoes]
Whatcha see? (Nothing
Nothing( Can you see it?)
Can you see?( Do you feel what I see?)
Do you feel what I see? (Whatcha see?)
[Verse 2]
Let me be your eyes,
For the day,
[Slowed down]
I'll show you,
(Let me show you)
[Levity]
What I see,
When I get to look upon,
Your Lovely Face,
Everything your troubled mind,
Might miss,
What it all looks like,
Together
Joined together in our,
Harmony
[Chorus]
As we deepen our connection.
What can you feel with my palms,
On your beautiful face?
Can you feel?
That,
I love you!
(Love You.)
[Short instrumental break]
[Verse 3]
Take a rest.
I'll do my best.
To help you see(!)
[Soft Echo]
By feeling with me.
[Racing Crescendo]
Feel our heart beat,
Feel our pulse,
[Whisper]
Feel them still…
[Very Short Pause]
[Shout]
Feel them race!
Feel my mind,
Know my Thoughts(!)
[Verse 4]
[Very Slowed Down]
Know that I do these things for you,
Voluntarily,
Unexpecting, Of Perfection.
Never costing you,
Your convictions, Your beliefs.
You can expect.
[Verse 5]
Unyielding!
Unwavering!
Devotion,
And Loyalty,
Before my,
[Loud Echo]
Vices!
[Outro]
[Very Slowed Down]
[Whisper]
You can hold out your hands for snow.
It's like holding falling stars,
In the flat of your palms.
They can only eclipse,
The Beauty I see,
In your teary eyes.
When yours, meet mine,
As my hands fall away.
And I say.
"It's your turn, tomorrow, to see how I see things."`
I really wanna know why the lyric video is breaking. >.<
Unfortunately nothing we can do here. You can report the issue to #🪲┃bugs-web
In the past, when I had parts that disconnected, it would normally be due to hallucinationa. I believe inserting text of the hallucinations (and removing the song structure) then aligning the lyrics in edit mode might fix it
Generally, it is not worth your time trying to get the lyrics video working correctly. Many people just place/generate a simple animation for their music viseos from sites like vizzy.io or go full out with other generative tools like SORA and Kling for animations
Anyone is extending songs with extension following the original version as it should have? What happens I can only describe as **extension is never created with same "seed" as original. **
It's the same result as if you "reuse the prompt", except that it adds nice transition into new part.
https://discord.com/channels/1069381916492562582/1308499132356362383 Please consider to make the stuff posted here at least "read only" if you decide to close it for posting. Right now you can only see the topic, but nothing that's posted in those threads.
How can I alternate verses between a female rapper and a male lead vocals?
Are you trying different Extend From times? The other thing you can try is using a Persona while Extending, that will lock down the music style more.
my question is that now thats up to 8 minutes is supported how can i force suno to make my songs longer?
when it was up to 4 min it was using the full length pretty well but now i still tend to get 3-4 min songs
mostly i would want it to extend the musical parts not just repeat lyrics many times
So, any tip on how to make the Persona create new music instead of creating the same track over and over again? Now when v4.5 had the ability to create great style prompts, why doesn’t the Persona function work at all???
My 4.5 Persona keeps recreating the original track:
- Try alternative Personas made from crops of the same song. Specifically try shorter Personas, 60s, even just 30s, of different sections of the song. You can also create a Persona based on the "new" repeated track instead, just Crop it first. Even though it sounds the same it may be less stubborn.
- Mix in shorter Covers of different songs too (15-30s) - unpredictable, but will mix things up. (Personas can be used while Covering now)
- Extend from a very short (3-15s) of any other Suno song instead of starting fresh. (Personas can be used while Extending now)
- Make Text Prompts More Different. If the new song uses similar song styles, change them more. If the new song shares a similar lyric style, use throwaway extra verses in front or after your real lyrics that are more stylistically different (you can remove these parts later.). #4 is not easy to explain in a couple sentences, I'll try to write a more detailed answer later for this.
- Use the new Creative Sliders and change Audio Influence in particular, but all sliders can influence Suno to be less stubborn.
One of features which I found in v3.5 and didn't in v4(maybe just bad prompt) - it's creating a narrator's voice, why that?
Why create bottom dont work in app?
Does anyone know if thumbs up/down feedback persists for tracks that I delete immediately after? i'd expect it to, but just curious to confirm
That was very helpfy. Thnx a lot 🙏
I'm not sure. But if you asking about the effect on your own personal songs, it doesn't matter because thumbs up and down are only used to improve the model for everyone. Suno has said they would like to someday let users influence their personal music with upvotes, but it doesn't do that right now. So it doesn't matter except in the sense that the votges influence music for every Suno user.
My personal opinion is that if Suno makes upvotes change the music for just me, personally, I will really want to be able to create different user profiles with different votes. Though as long as I can disable personalization it would be okay short term.
I know some other music AIs already have personalized generation with upvotes. Curious how it's working out!
Yup, I'm asking about whether its going to help curate and improve the training data for a future finetune. I've got a pretty niche usecase that every single model until 4.5 has consistently failed at - and this new model only works maybe 15-20% of the time (and degrades pretty fast). Hopefully if I contribute enough data for reinforcement learning v5 will finally pull it off
I think Nvidia Fugato or StableAudio2 might have a higher success rate at morphing vocals, but theyre inherently worse at music so thats the drawback
Degradation my guess is a pretty fundamental or technical problem, but there's so many weird tags and niches, doesn't seem impossible a single user could nudge things.
I'm nervous about Suno leaning too hard on voting for global changes. Thinking of stuff like the recent OpenAI "sycophancy". https://openai.com/index/expanding-on-sycophancy/ But for personal use, a vote could be great, like a hot or cold signal to hone in a style.
That said Suno has a long asked users to vote, especially downvote, they must get a lot out of it.
Pretty sure we could pull it off - a few thousand +/- annotations for 3-4 niche tags will constitute a significant (or at least sufficient) portion of the RLHF, most people don't vote that much
Finetuning works well with far fewer samples than that, both for LLMs/Diffusers and I'm assuming Suno is also using a DiT architecture like Riffusion migrated to with Fuzz
Is there a Fugato demo? Or just samples?
looks like only the paper is out now, but inside it mentions they plan to release the full code and dataset
is anyone having trouble extending I keep getting credit refunds and no extension to my song
I got the same problem with 4.5. I’m Getting same baselines, riffs and sometimes even lyrics bleed over from personas
Same. I think it's broken. But then you get a thousand people telling you it's not. It has t worked for me since the 4.5 update.
Yeah, but I will try what Jonathan says about sporter clips when creating the persona.
" Bummer. Its not us, its them
I've been trying to generate a track with no instrumental. it seems to be very difficult with 4.5.
here's an example https://suno.com/s/64HJUedfukeAVnAp (using a Silence persona to help, but it's still producing music)
another example https://suno.com/s/iLjKOWgefkieBz0V (without persona)
4.5 does reproduce melodies (and voices in Covers) too often in my opinion as well. It's not all the time, just more often than v4. The important thing is to change things (like in #📚┃suno-school message) and you should be able to get something different.
Extensions are also more often short. This is similar kind of thing, it's just more often than before. And if you change something (Extension Time, Lyrics, using a Persona while Extending, or a different Persona) in my experience you can always Extend the song. I don't mind this problem much, except when Extensions are short but not short enough to refund credits.
Is something going to be anything added to bring back the ability to choose a different length with section replacement. I've currently got a song that I like but the guitar solo is way too long. I love a good guitar solo, but this song isn't the one for it. I'd like to shorten the section. I tried removing a chunk with the intent of using section replace to cover that up, but I keep getting an error if I try to edit the song after I've removed a section. Hmm. I have an idea, but I would still like that ability brought back.
Hi Suno Team,
I'm reporting a persistent bug. When generating longer songs (around 4-5 minutes), often requiring the 'Continue' feature, the AI frequently skips large sections of the provided lyrics.
For example, even with correctly formatted lyrics for an A-Z song, it might jump directly from the section for 'L' to the section for 'U', completely omitting the lyrics for M through T. This happens consistently, even after trying shorter lyric versions. The input lyrics are within character limits.
This makes it impossible to generate the intended complete song accurately. Could you please investigate this issue with lyrics being skipped, especially during longer generations or when using the 'Continue' function?
Thank you for your great tool.
For some odd reason my Suno refuses to generate any extended audio with a voice of any sort for a new song I'm working on. And i've made sure to turn it off Instrumental
Is extend works or is it still not working???
It's been broken for me since 4.5 came out.
Ya me too grr
Lol i tryed to download a audio with jungle sounds and it rejected it by saying they spottet vocals in it ????
hello i have a question, how do we use the smart prompts in the new version 4.5? I want to transition from one genre to another, do I explain this in ''Style Description'' or in the lyrics part?
Try do it in lyrichs alot more space for it
But just be aware that the collums has to be set correctly or else...
When i prompt i usual do it in lyrichs and the make another prompt for music style that combine the prompts from lyrichs
Can anyone tell me why a song I just generated is now not available in my libary? I just made it and now it's gone
can someone help me please with extending songs because the lyrics are just highlighting red and not generating extensions
does v4.5 actually recognize chord progressions? saw this posted on reddit but i dunno if it actually works
Though this might trigger a single random chord change or a modulation in general, it is at the least, overkill.
But let's see over a consistent use and 100 samples of equally consistent results. Then I'll change my mind.
yeah i figured, maybe at the least you can specify what key the song is in but doesn't seem to follow much more granually than that
D major
I hate 4.5 lol it's kinda ruined SUNO
Most of my new songs don't exist and haven't shown up...they disappear and never come back after you leave the page
Hi everyone, sorry for this question, it seems that songs created with the free tier can't be monetized on platform. I subscribed to pro but is there a way in our suno account to identify the songs we can commercialize and the one we can't?
There is a odd bug, that takes my new songs away...they won't play but they only show up on there personas
It seems a glitch. It seems universal. And it seems indescriminant.
As part of the beta, if you can repeat anything which seems to trigger that, please note it specifically in the bugs 🐛 channel. Thanks.
Thank you and okay
Does anyone kow how to get scores for the music and the vocals? thanks!
Yes. And personas, and some say, covers, are tending to do this. Helpful to know that you can see similar when you have the same issue.
Sorry for the trouble though. It's not a good thing, for sure. Let's hope it's easily rendered gone soon.
Since this seems to be happening frequently to you, can I just get a rough overview of the process you use when creating these songs and what actions you typically take between pressing Create and the song disappearing?
For example the way I work, I press Create and then I wait for the 2 songs to be playable, then I preview a little of each by clicking on the artwork of the song in the song list and I might hop between the two.
Do you do anything different, like are you going into the song pages whilst they are generating, editing the displayed lyrics, adding a caption, anything like that? Basically anything you do with the song whilst it's still generating.
The persona voice timbre is not working from several days ago please HELP😂
If you've just upgraded to a paid subscription, the obvious answer is that everything before that time of upgrade was a "free" derived output.
But from that point, modifying any "free" song, there is no means of Suno tagging it for us as permanently "non-commercial use".
You may have the benefit of tracing the outputs through it's history as long as the history lineage shows up in every step of our modifying a past free song. Or make sure you place them in a specific playlist of non-commercial Free songs, or before working on one , note in a workspace that this was a FREE song.
Thanks ! So easiest way is just to create 2 Workspace (Free & Pro) and Take the creation date as a marker ? Also I can still publish my free songs if I don't turn on Monetization on YT right ?
Question is it possible to make parodies like monster from skillet in Suno
100 samples of equally consistent results.
... I don't think that can be obtained for anything produced using a generative model
This technology is fun but unreliable by nature
I say that but I was telling someone this week about the style tags 8-bit, 4 Hz to obtain a videogame like song track. And the fact it's consistent, that's one of the very rare cases I can affirm works consistently with a generative model
".....if I don't turn [it] on....."
The wording in the FAQs in in the affirmative instead of the prohibitive. It says, that free users have permission "....Using your music on other non-monetizing platforms...."
So thats up to interpretation of the legal definition of a "non-monetizing platform".
TOS notes: Commercial Use: Subject to the Content Section below, unless otherwise expressly authorized herein or in the Service, you agree not to display, distribute, license, perform, publish, reproduce, duplicate, copy, create derivative works from, modify, sell, resell, grant access to, transfer, or otherwise use or exploit any portion of the Service, and any Output, for any commercial purposes.
Between the two, you be the judge. A good attorney can be helpful. 😉
hi so im having a problem with replace section like it dosent show the lyrics
Hi see thanks a lot !
Yes. One version of the ".... 100 samples....." is that 100 users do something similar once (and discriminative - a single or only a few prompts, not buried in several lines of metatags) that we can see some consistent and positive results over a period of time.
I'm happy some things work. Over time, they become standard hacks.
One person, posting some elaborate concoction of tags on reddit, with no audio even, is just trolling vanity and sus until proof otherwise. No need to repost it here and ask "is this true?" . 😅
right, 100 users producing a sample with the same params. that's another way to collect 100 samples.
There's indeed a part of it being a blackbox; we don't know how the platform is configured and how dumb the system prompts are... as an example, the date/time + user real name and company name being added dynamically to the system prompt in some subscription tiers of ChatGPT... That's so ridiculous I can't even imagine they do this.
I demonstrated how, without telling anything about Valentine day, the model might start talking about it just because it was Feb 14.
I finally upvoted a song and it opens with a specific "aaaaaah ah aaah ah ah" (no idea how to describe musically) sequence. Now all my generations are starting with that sequence unprompted. Will it go back to normal if I remove my upvote?
That's new. 🤷 Try it. Remove the vote and run a generation. If nothing. Put the vote back in, run another generation to see if it comes back. Report back.
Why just the beginning "ahhh ah ah"s." Why not the middle riff, or the drum pattern? If it were true, it would seem every one of your generations would have more than one element of the previously liked song, and change consistently, every time you up-voted another song.
Certainly a wild coincidence, but if we can repeat it, then that would be quite interesting. Let us know what happens.
I recon my method I could poop out 100 songs of consistent quality
but like you mentioned that consistency amounts to generative ai sound
Where is option to change song cover in the new UI? i really cant seem to find it.
Using android app and mobile version of the site.
song details in the ... menu
S/O to everyone being thoughtful and helping other people 🙂 Always a big deal!
@daring lava my friend I have more songs capable of blowing minds
pressing cover does nothing
if you ever need entertainment
It's becoming prominent that certain quality and fidelity tags are helpful. The beauty of an ever emerging generative AI.
yeah this thing does what you ask if you ask right
with clarity specificity and structure
I will investigate mine see if it is the same
Thank you
I don’t doubt that one bit! Will be on tomorrow to check out what people made over the weekend! Never hesitate to add a ping to a share, I’m pretty good about circling back around to catch up on things!
do you have that when you do to the song menu on the app
with the pencil over the corner of the image
i wish i had android to know precisely
Does Cover show back as a "beta" form again? In app? Not showing beta in desktop (least i think not).
i used to but now its gone
ok let's go back to the site we should get the same site, i'll look there if anything changed
on site it's working now, just tried it.
ok, i'm sorry the android app seems busted
app development is significantly more frustrating that web development and my understanding is they're trying to hire more mobile developers, you could go to...
and make a thorough post about what it was and how it's different now
i can write dumb ol' java maybe they will give me the source 🤣😭🥲 (unemployable)
OK. New and funny problem. [The following is tongue-in-cheek]
When making a J-pop duet, those darned males vocalists like to sing in falsetto and hard to distinguish between the female vocals. Dang it. Not good for the musicality factor. 😂
Most male Asian vocalist have higher register these days, Im not surprised of your results 😄
I like some jPop male groups having that one sultry androgynous bass vocal guy who sings the one-off ballads. Its all a thing, but dang hard to write for.
Maybe? They are generally more slender people, so there may be some merit to that.
This primarily applies to remastering and when extending an upload. Instead of shimmering, we now get some kind of metallic/tinny sound, most prominent on the drums – especially the snare and hi-hat. It sounds like a plate reverb with zero decay – basically a plate reverb, but completely dry. It appears after a while and gradually intensifies until it’s at its worst by the end. Simultaneously, distortion kicks in alongside the metallic noise. You have to wonder when Suno will release a version without these audio glitches.
It would help a lot if you provided some direct feedback... go here https://sunomusic.typeform.com/to/hcFMxd2L
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
A questions about the style tags in suno. Are these styles like hardcored styles which you can use to guide suno into a specific direction or are these just suggestions ? Because i have created a java script and i am pulling all the tags from version 4.5 from the system and i already at 3040 unique styles and some i even never heard of or seen like : 3d audio , festival anthem, perfect eq, Full Song etc
suggestions
you can write to the AI in the style box if you want
4.5 is surprisingly good at picking up what you're putting down
but also some tags are silly or asperational
The style box is now more responsive to complex description of styles... and this is overly evident when you type in your basic styles and hit the enhance button..
As well, the negative box has a lot of weight
It's just a bit annoying when you enter piano the singing voice is automatically softer and suddenly sings sadly in a strong rock song
I know but i want to know if the suggest style tags are just a helping hand or actually official tags used by suno to get your song in that direction ?
Describe what you expect in teh style box
ai is not programmatic
there is no list of commands
full stop
Hmm, how about trying some "smooth male vocals"? I find that typically gives me more masculine vocals
Though I like how the magic prompt translated this:
- Smooth male vocals --> warm, mellow male vocals
- Cheery perky girl vocals --> bouncy, bright girl vocals
so these style which appear are just randomly created by suno to program the ai ?
oh i have do that multiple times 🙂 but nothing work. I Write " Explosive Rock and something else but Piano killed the power in near every song
no idea, but there are some goofy suggestions sometimes so I assume it's based on recent user popularity or something
@smoky crow you will never get it if you think in terms of programming
it's a complex set of linear equations for which you are given a stochiastically controlled randomized output from the solution that was derived to make the model
it's not a programming language
over time our actions and team actions may carve that solution space into tag influenced spaces, but it's not programming
and the outcomes are not guaranteed
Here is an examnple of what I had to do in order to get my latest song to structure the way I needed,
A slow Irish folk ballad led by a bright female vocal, captured with high-end condenser mics, Sparse acoustic guitar and gentle fiddle intertwine in the intro and verses, joined by warm tin whistle phrases, The songs opening is spoken like a poem and then joined with an instrumental into, After verse 1, a lively double jig with bodhrán, whistle, and fiddle creates energy, before settling back, Verse 2 returns, then transitions into an energetic reel with full ensemble interplay, The arrangement alternates between tender minimalism and lively traditional interludes, ‑heavy reverb, ‑cathedral-style reverb, ‑long echo tails, ‑excessive vocal reverb, ‑muddy ambient reverb, ‑over processed reverb FX, ‑distant-sounding vocals, ‑ghostly vocal trails, ‑washed-out background layers, ‑reverb-heavy snare, ‑room echo, ‑ethereal haze, ‑dreamy vocal spillover, ‑spacey mix down, ‑overly ambient production
AND, I had to use meta tags in teh lyrics to reinforce the places I needed specific things.
Yeah the style suggestions as far as I know are just what other people have used. I've seen some with typos.
ok i can understand that but i wanted just to know if the styles presented are official tags or just a hint you can use so when you start creating the music you will give it a direction. and these styles are not part of the program but a suggestion
There is a predefined list of generic genres the suggestions is pulled from, or at least was the case in the past, but the AI has learned many more patterns than that list would provide.
I suggest looking up a list of subgenres so you can get the type.of song you want then use additional keywords to fine-tune that.
I personally prefer https://rateyourmusic.com/genres
it's a suggestion
basically communicating to the ai what you want
and hoping to god it doesn't make some random nonsense
I wish SUNO gave us the option to Edit the Audio Highs and Lows; Or just fix certain things all together.. The Muffled audio sound on V4.5 is annoying, but at the same times the songs sound better than V4
That tends to be what I aim for sometimes 😄
and you've been answered 🙌
i have burned my 10k points away in 1 week xD i hate and love 4.5 .. 4.5 cant cover songs.. the cover in 4.5 are very bad.
that's wildly against the rules
ok?
fine me then
or do you mean through the browser
through the browser
Example with one of my songs; You can tell the sound is muffled, can barely hear the instrumentals - Like their being covered by a pillow lol
not the api
YA, thats been a wide ranging complaint among other big problems.
created a java script and that pulls the data from into a file
i am confused and wonder if you're afraid to use tags you don't have official certainty about
Nah, they won't fine you. Chances are your account will get restricted upon detection of usage for this type of thing... so thank you for letting us know what to pass along to the devs
yea but! When you make an own song its bombastic. I have make Limp bizkit like music thats so nice 😄
i think he over represented what he did
1940-45
he didn't pull API
he pulled from the local cache
no one is getting account banned for reading local storage lol
Is there a particular reason?
Ah, if it was local storage, yeah that's nothing. Thought they were hijacking the j-query process
Its still a new model so we are all still tryign to learn what makes 4.5 go, and some of us are having better luck figuring it out, but its not cheap. Ive burned almost 2000 credits experimenting and learning
This is what i used in the browser when open up with f12 and then use console and click on a style it will collect styles into a file
thats true 🙂
I see. When you said you got it through the API, I thought you used a more intrusive method rather than parsing the list through the inspector tools
I literally battled Suno yesterday for 13 hours trying to get this... https://suno.com/s/PwABWOEFEYRGkaai
Suno for hours upon hours refused to give me the structure I wanted, and what I ended being successful with was a combo of detailed style prompt and meta tag direction. I was exhausted by the end...
Heavier Metal/Rock songs are the one with muffled thing, but I see that something more slower pace the highs, and lows are good
2 things i really like now from suno 4.5 is the extended lenght and the smarter prompts. now when i type in techno i really get techno instead of k pop mickey mouse house on speed
Right? xD
now I want to try mickey mouse metal
Yes, we noticed that immediately on Thursday when the model dropped, I also write Metal and it was obvious in 5 seconds.. Right now I am not doing any more hard rock or metal until that beast has been slain.
The only thing i would really like to see if we could say create a song of 6 minutes when prompted to for 6 minutes
Well, at least there isn't a shimmer
I'm probably going to run my songs through an audio enhancer program, before I upload them to Soundcloud
YOu can, I have gotten 6:50, but yo uneed enough content in the direction for Suno to make them that long
Not for instrumental music
Ya but the entire band is playign from the inside of a mud puddle 😛
This is the best Nu Metal song I've made with it so far - The highs and lows are perfect for this emotional of a song
For those who are interested here is what i got so far from the style tag list
hi people, tell me, I haven't been able to create a track for 3 weeks, he writes that couldn't create due to internal error, please try again later
what are you prompting it?
Instrumentals are no different, you just need to use different tactics now to add length... One user figured out that using punctuation as lines in lyrics makes suno add length
Are you banned, or something?
In instrumental how do you that ? I don't use vocals.
Also, Sumo is still largly ignoring BPM and tempo direction, overriding to what it things the style normally has
its like
,,,,,,,,,, ,,, , , , , ,,,,,,, ,,,,,,,,,,,,,,,,, ,,,,, .............. .. . . ........... . ... !!!!! ! !! !!! !!!! !!
I've seen the result, it actually worked.
These are all the style tags to improve your sound quality when generating tracks
try making the instrumental in the custom mode
it's not that they banned me, the account is active, I just want to create a track, writes that an internal error has occurred, try again later, tried from different devices, VPN, a new account. I just stopped creating, both on the new version and on the old one.
did you update the app?
I asked around in the fan chats, they say that it's possible that the servers are overloaded, but I've been doing this for 3 weeks now
Yes
yeah they have not been overloaded for 3 weeks
you've tried mobile app only? what about mobile browser or pc browser?
When extend a song . How do I choose the right song to mash my type of song that I chose
I am trying to extend a song, but it seems like, as others have experienced, to not be working properly. Song was created with v4.5, and then extended with v4.5, giving only 22 seconds of quiet music, virtually nothing, not singing the words I put in. I tried again with v4.0 and got 1:30 but still not singing the words I put in, and not really making much music.
are you extending from far enough back and not end of song?
From about 80% of the way through
The same problem exists in the browser and on android, but unfortunately there is no way to check on a PC.
Ok. I need a new career. 😢🤣
are you clearing the lyric box and then adding what you want to extend? If you leave all the original stuff in the lyrics box, Suno tends to ignore it. You need to use normal meta tags and then the added lyrics only, and also make sure you have the sytle prompt in tact.
Oh, miracle, it worked. 👍
So I tried again, and went back a bit further to extend from, and got a much better result.
curious
Sorry, I tripped over the plug. Plugged it back in for ya. All good now. 🤭 (j/k)
Is there a way to prevent the AI from backtracking to a previous verse?
I've tried specifying verses, verse numbers, not specifying at all. but it will ignore verse 3 and repeat verse 2 after a chorus/interlude. 1/10 generations did it correctly
Seems like my tags for audio quality delivers. for the 1stime i have now 4 tracks without the bloody compression
This is on a new gen ouput? (Not cover, persona)
Persona voice timbre is not working from several days please help
Ive experienced this, what I think 4.5 is doing is overriding what you want and doing what it thinks is popular from its training...
Are your lines in your verses similar in length and syllables?
My trying to avoid saying and advising "extend".... I can suggest adding additional non-descript tags such as [pause] or [clear] on a line of their own or even give a couple of line spaces, along with slightly altering the lines of each stanza to alert to a different segment of the song.
It seems to help if you have "just enough " lyrics, which i think about 2/3rd down is a sweet spot. Too few and it (used to) tries to lengthen the song by repeating verses. Too long and it can repeat a verse just to make things easier than to add additional information. All speculative , of course.
They are similar length and syllable
Since you have lyrics, you can also add [male vocal] before the verses to signal that it appears to be another verse, and without anything further, hoping it uses the same vocal as before.
might try that
[Verse 2], [Verse 3], etc. did not work. I'll have to try tonight though because I'm out of credits
I clicked on showcase instead of technical and about had an aneurism that it had been taken over
We try to delete all our song posts when we know you're on your way in, but aren't fast enough sometime. You caught us.
Hello, I have a question, how do we use the smart prompts in the new version 4.5? I want to transition from one genre to another, do I explain this in ''Style Description'' or in the lyrics part?
Their promo seems to be geared to the most simplistic means, by using the style description and it was exhibiting an instrumental, so it could be even more interesting with vocals.
Certainly, it's conceivable though that proper tweaking can having it do similar change ups in Custom node as well.
Who can help me here when I got a premier plan and I get this option greyed out when I TRY to create a persona ?
.
Pizza?! Woodhouse say pizza? I'm there!
Certain parameters won't permit personas. Using an uploaded audio which has vocals is one. Perhaps that's what you're trying?
ye that could be it, should it be more evident on hover that this option is 'greyed out' because of that and not because I dont have a subscription( as it says now ) tho? 🙂
also its my vocals, how can I get around that?
You have a Prem and it says you need to upgrade to use persona?
this is what I get when I hover the mouse ( lost my mind haha )
did check and recheck 😄
I can't say I've seen that before. New UI notice.
The web UI is still full of bugs
Please report it down in https://discord.com/channels/1069381916492562582/1082775596767117312
coolio, also when I upload my own vocals how do I create a persona?
do the devs look in here? Where's a good place for feedback?
They have that disabled, you cant persona vocals you upload.
well thats a bummer 🙂 thx tho
Look in the bottom of the channels lists, in the FEEDBACK section
Yea. So, you should be good to go, just can't persona against a vocalized upload.
cheers for your help mate
I don't have permission to post in there, I've got an active account.
Can't post here ? >>> https://discord.com/channels/1069381916492562582/1082775596767117312
Or suggestions as feedback https://discord.com/channels/1069381916492562582/1154587937611190354
I can post in there, not under v4 feedback
Yea, right. That's just a place card... the sub-channel is for V4 language issues.
For 4.5 feedback use this link https://sunomusic.typeform.com/to/hcFMxd2L (fixed)
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
ok thanks
I wonder if 4.5 is doing better for all languages now.
Its certainly better for Gaelic, i've heard complaint though for Romanian and other Asian languages
Still haven't tried 4.5 myself 🙂 soon....
Part of teh problems with many Baltic and Asian languages is the dialects, I believe.
well, good the update has a positive momentum then for languages. I'll have to see how it likes to do accents as well. no one gets Australian down well, so far.
Ya, there seems to be varying depths of how much the accents vary from area to area from my experience... there really isnt just a one accent fits all in Australia, much like the US has 100 different accents for the same language
OK. New Zealand? South Africa?🤔
filled out
Ive done a half dozen of them so far
The persona voice timbre is not working please help
the message that is shown on disabled options is always "subscribe to use" when it is disabled, regardless of the real reason that also could be things like "disabled because you try to make a persona from an upload" ... i reported that it behaves weird.
No staff monitor this server. You have to email them.
Ok future request. Once you have uploaded your own audio and you want to use that same audio again, please let us just pick the already uploaded audio from a list instead of searching for the file or reload it again from a local machine
there is a suggestions channel, that looks like a good one
Just like a persona where you can just select onem they should do it as well for audio
timbre is not strictly confined by the persona feature, you may get the same, but not always
I also noticed that when you cover a song created with 4.5 with another 4.5 version the eq is turned down. a workaround i found is when you created a v4.5 track use the v4 cover option
I m trying to download my song from suno as usual but can't succeed as I should, help please.
please use the website . i guess you try to use the android app? this doesnt have download yet.
Or is the v4.5 can't d?
you can download everything that has generated properly.
Ok
Can't seem to get a 303-style acid bassline out of 4.5, do you have any tips?
No i always download through the web , only today i can't download
if i cant get a version to do a specific thing, i make clips in the versions that can do it and than i cover it with the newest version and the same styles. but maybe 4.5 just needs more specific descriptions. as funny as it sounds one of the most interesting 303 variations i got by using the didgeridoo... https://suno.com/s/xdUQze1gznaAVxUU
can you try another browser to test? can you download old songs?
Cheers. I will try. I guess recreating the variations in filter, resonance, and distortion may be a step too far.
you can always use upload... https://suno.com/s/bWNIKufNPN3ND4wG
the thing you hear in the beginning i made in my daw and just let suno extend that...
Its will try
hopefully they will fix the cover mode in 4.5 because it pretty much useless right now. it's like the sound gets a pinch down in the upper area of the high eq spectrum
the lyric fixer needs some work. We should be able to select audio snippets and manually assign them to lyrics ngl, because the model is really bad at it
You definitely should suggest that at #💡┃suggestions-web where the devs can see it
hello. I want to ask just in case. if I click on the trash bin, will it erase all my library of songs, or will it show what i moved to the trash?
It shows your trash folder
i know it's kinda of a stupid question. thank you
You welcome bruh
any way to upload an orignal song and make it public?
Also, its a mixed bag, it will let me share some non-lyric uploads, but not others. Suno did an amazing job remixing one of them and anxious to share.
I'll give this a go, cheers. Although I've tried on different devices, still not there, also I got the credits back so I'm pretty sure they're gone forever 🥲
Looking to improve the instrumental stem and get rid of vocal artifacts. So far, I get the best results by taking the instrumental track and regenerating it again with 4.5. it seems to reinterpret the vocal artifacts as some instruments.
Okay I guess this is what the problem is for the disappearing tracks. Make sure your using actual genres
It's all about if it has vocals in the upload. Your first or so uploads pops up a notification stating as well that anything perceived as vocals will limit the output (and any dervatives after that) from being made "public" or "published " now.
You can click and copy the link and share, using the "box and arrow" icon.
Does anyone know why V4.5 automatically cuts my song off at 4 minutes even though there are still more lyrics, or why I cannot extend a song? The extend function does not work in V4.5.
Is it a hard cutoff at 4 min? Verify if it was actually genned in 4.5 because I had no issues reaching like 6 min (albeit those involved classical music).
If it naturally ended at the 4 minute mark, then that is within the norm - most people do not want songs that would reach 8 mins so the AI does tend to end it early unless there were plenty of commands or lyrics to extend the generation time (e.g. instrumental interludes).
Very surprising to hear Extend doesn't work, whar happens when you used it?
I even need a 30 second jingle now and then. 🫡
It starts to generate, then stops with a note popping up saying "credits refunded". This has happened the last 23 times I have tried to extend my song so that I can get the last two verses in.
It just won't extend the song. I don't know why, and I'm frustrated.
Oh, and when you were extending, did you extend at the very end where there was a silence, or did you go back a bit?
That could be the reason why, the AI was likely in the terminal marked territory. Try going back like 5 seconds or where the music was still continuing strongly (a few seconds can make a huge difference)
It worked. I didn't even think of that becasue in 4 it would just start at the every end.
Thank you for helping me figure this out because the last verse is the punch to the song.
While each model are very similar to the previous model, they did learn differently so sometimes you have to do things a bit differently. That extend is a good example.
Another example is how you could [End] in 2.0 and 3.0, but the models "forgot" (or did not sufficiently learn?) how to stop after them
[Fade] now works superbly
So is it me? Or does 4.5 sound muffled compared to previous models?
Common complaint
Certain genre are really muddy and muffled without specific direction to NOT do that
I hear it mostly with remasters..
Remasters and covers all suffer the same issue... you can help it by using the negative window
Especially remasters from 4-4.5
all of my 3.5-4.5 remasters have been magic
I see. Then I'll just have to fiddle with everything until I find what works.
I've had many sound muddy on regular playback and sound good in the playback in the Edit UI. Very odd. I'm planning on downloading those firstly to see if it sounds any different. It may have just been in a day with a Y in it. (or everyone using 4.5 last week. 😀)
I already went back to 4.0. Everything so damp with 4.5
Is it only me or does suno somehow prefer female vopices?
Just another tip i want to share after my analysis of downloaded style tags if you want to push the ai to create 5 minute + tracks add these tags to the ignore list : 4 minutes, short
I have investigated this issue and you can get better results add this to your style tags : lossless audio, long track
How do you put an exclusion with a Remaster? Try to cover it first?
Thanks. Good suggestion for new gens. We were talking about Remasters and I didn't put that detail in my response, sorry.
Depends on the genre. For example, Metal heavily favors males while pop mostly favors females. Girl Group (yes, an actual category) would be all girls theoretically
You can skew the likelihood to the gender you want by having more weight by repeating term sets, like "male,male,male" or "male,man,boy" to favor male vocals a bit more. Using the exclude tags helps a lot too
Here is an example (v3.5) where I was able to have a Girl Group with male vocals, though I only focused on the metatags and it was like the only track of six I tested with
https://suno.com/s/iCDmASpmoyCOoNB4
Girl Group but influencing the vocal style through the style prompt only, no excludes, 3.5:
https://suno.com/s/lR1qA6j2o0C8YI9R
And here is a 4.5 using the magic style prompt to update the style used. No excludes
https://suno.com/s/4YJLf7S5oJJG8ayy
On top of this if you try to make metal with female vocals it tries very hard to add pop flares to those songs, and sometimes overshoots into just forgetting all the metal prompts at all. It's always done this at least a little bit. Sometimes the mix really works, but lately not as often.
In v4 I was getting mainly female vocals, in v4.5 it's mostly male vocals. This is when I don't declare a vocal style.
create a female metal song in v3.5 and then create a Persona from that, it'll carry over the vocal style into v4.5 quite nicely.
What I describe is true for all models and has become more true as time progresses. I already do have personas from several months back thankfully, but this has been the case for at least six months
But 3.5 is at least less bad about this
I'm making metal tracks in 4.5. EXTREMELY high in the 4k range.
Why does Suno take (S.A.T.B.) as a name when we put it in the prompts, when it is the acronym of (Soprano, Alto, Tenor, Bass) 🤔 🤨
Just an observation: (and I think a cool one) 😎
In over a year of listening, regardless of versions, I've heard at the end of playback a phenomenon that's makes me think of "tape drag". Where, especially on a held chord you can hear either the pitch rise or lower slightly. I can only assume, depending on which song style Suno is using, it is literally emulating some of what its trained on? Analogue recordings where there were fluctuations in tape speeds, especially near end of the reels or a pinching tape drum. Ive heard a couple of times inside a song, but most often its at the last few seconds. Here's an example i got, where even a Remaster thinks it's normal I guess-- which it kinda is. 7 seconds sample.
how are you guys adjusting to the way your personas sound in 4.5? Mine sounds unfamiliar. I mean it’s great, but I really liked my 3.5 persona. I got it to sound very similar to me, with my tone and range. This new persona is like going from an average guy voice, like Staind singing It’s Been a While, in 3.5; to System of a Down, singing Chopped Seuy in 4.5. It’s not just an upgrade, it sounds very different.
I'm noticing when I use Persona, they often use the exact same song and it just changes the updated lyrics. 4/5 songs are exact same intro and overall song
What's the point of (Exclude Style) if it doesn't respect the exclusion of voices we don't want???
As far as I know, the vocal labels are more implicitly learned in regards to gendered vocals rather than explicit supervised training, which would explain the past ~60% successful rate for duets I had in 3.5 (well, successful in getting both genders, though half the time it swapped the vocal lines...).
If you want to improve your chances, skew the term weight to the gender you want (and want to exclude) by repeating it like three times with commas in-between, like "male,male,male" or "male, man, boy", etc.
This does not guarantee that you will get the vocal you want but it does nudge it a bit more
I do it 3 times "Female Voice", and "Female Vocal" to be sure that it doesn't bother me anymore. In the arrangements to have male soprano voices, I put "Male Soprano Only", but no, even with that, it gives me voices to do, even when I want mixed male choirs "Soprano, Alto and Baritone".
Like I said, it nudges it, it is not guaranteed to give it to you.
I honestly don't think the AI knows what soprano, alto, and baritone refers to, as it never worked out when I tried it in the past.
The terms that I use that I know works are often descriptors like smooth, cheery, bubbly, etc. But that could also have been a coincidence 🤷
That's not why I stipulate "male" before each voice because there are also alto and soprano female voices.
He even managed to put a female Alto on a Tenor like here “[Verse 1 – Strings, Soft Piano, Male Tenor Voice Only]”. There is a problem with their data bank!
they kinda do the same thing (kinda) only with cover you have access to the prompt windows..
Something strange happened with Suno 4.5. I switched genres, and suddenly almost every track I generated sounded incredibly crisp and clear—so good that I saved a lot of them. It felt like I was in a perfect creative flow. But then I noticed my credits were running low, so I refreshed the browser. After that, I used the exact same genres and styles, but every new track I generated sounded dull—like the high-end EQ was missing, and the kicks were muffled. I made around 40 tracks after that refresh, and none of them matched the earlier quality.
That’s when it hit me: Suno likely uses render farms with many GPUs, and tracks are probably load balanced across different servers. Could it be that some of those servers have issues? Maybe certain servers behind the load balancer are underperforming. It was just so odd how suddenly every generation had amazing quality, and then after a refresh, it all changed.
The difference between before and after is ridiculously noticeable. I now have at least 5 tracks from before the refresh that I’ve mastered, and they sound like professional productions. But the ones I made after the refresh? They sound amateur in comparison.
Nah, just most likely RNG 🎲
Devs confirmed (several times) that RAM/CPU usage does not affect the model aside from the speed of getting your output
'Match Lyrics To Audio' UI, sucks... that's about all I can politely say...
note to self: when quality is good, do not refresh 😉
The largest culprit aside from RNG is your prompt itself
For example, if you use Pop as your genre, well there are a lot of subgenres of pop it can pull the learned patterns from. You will find that specifying such details will often yield better results towards what you want
Its literally rng.. I have had gen cycles where one gen will be crisp and clean, and the one that genned with it will be muddy and dull...
This was totaly different. the quality was actually amazing. Almost everytime. The ones i had to get rid of where most effected by timing issues which i never had. I had to refresh my browser because i run out of credits. the refresh had apparently connected me to a different server
lol I just noticed Lumine changed her name agian!
I didn't changed the prompt at all. I spended almost 500 credits on this because the quality was really outstanding. crispy clear audio. i just hit generated and saved a lot of tracks
At least i got 10 tracks mastered right now and it sounds absolutely amazing.
As far as I know, Suno has a single central server. How your webpage connects to the site doesn't impact the generation, your data is sent internally to the pipeline needed to process such data. It is not like an MMO
But if you strongly feel that is the case, then just refresh until you get a good quality? 🤷♀️
I created several songs, but I cannot see them.
I work as an network engineer and these guys work with F5 loadbalancers
Suno has a couple of F5 loadbalancers active and behind them they load the balance the traffic from all users to different renderfarms
Cool, so what do you suggest the developers can do to help better the process? You can let them know at #💡┃suggestions-web (the devs do monitor that channel and the bugs channel)
I have no access to that platform
That is strange, there should be no permission blocking that channel. Try clicking it if it says No Access, sometime Discord messes up.
It is under the Feedback category
That one server you see is a digital traffic agent pointing customers where to go to to avoid a ddos attack on their own platform
Are you using a persona? I have heard several people complaining about this situation but we community members didn't figure out the common cause of it aside from just that AFAIK
Also be sure to report the issue and what you did (e.g. used a persona from 4.0 with model selected to 4.5, etc.) At #🪲┃bugs-web
Also they can manage broken servers by taken them out and repair them and diverse traffic to other render farms or add more to the loadbalancer enviroment
v4.5 in this example here sounds exactly the same like v4 on my system 
if not a bit worse because it adds a hiss
the best of v4.5 also all sound like very low quality or low fidelity
Anyway, now I know—if this happens again, I’ll reload my browser as a test. The whole thing was just so weird.
make it more obvious on creative radio stations to hear the music like the current track
calling the first two verses collectively [Verse 1] and having the next verse after the chorus [Verse 2] seemed to make it less confused. originally I had those verses called [Verse 1] and [Verse 2], then the one after chorus was [Verse 3]. also added the male vocal tag to one of them. thanks for the help earlier
Honest ? sounds like suno
Ya it is
I mean it has the watermark of suno on it. You can tweak suno to let sound like it's not a typical suno track. the force is within the lyrics
Hey, If it works..... 👍 No problem.
Suno songs are not really for "rating". Typically, IMO, the lyrics get about 95% weight and the remaining is for clever arrangement, otherwise I'm old school and respect the musicianship of the band of which , well......
So it's cool you got a poppy indie grunge style song and some catchy lines here and there. Feels like a constant 4-line , 8-syllable verses, but vary between freestyle and metered singing. The typical 4 chord progression popular since about 2004 with no change between verses and chorus. Always a go-to style for streaming radio but even with curse words, can become MOR style amoung the many similar. A few poetic lines that slow it down a bit. All in all, an entertaining song, lyric wise. Rating 1 to 10 would have no meaning. It means alot to you and so you should own it and be proud of the effort and time spent. Cheers.
who do i talk to about songs getting deleted that had nothing bad in them?
Songs don't just get deleted after generated.
Maybe you made it in a different Workspace by accident? 😉 Happened to me a couple of times 😄 @dim zealot
i only have one
they shouldn't but they did some how
anyone knows if it's possible to reorder tracks? If you move to another workspace they stay in order as you moved them. not in original order, it's a mess.
I've seen quite a few reports since v4.5 drop;.. that songs didn't show up after creating... but maybe a display issue
But if they don't show up after refresh then it's a bug.. check #🪲┃bugs-web for other similar reports or report it yourself.
oh i did and others are talking about it too. luckily i downloaded the two i intended on keeping
it is concerning none the less
Good. Hopefully they'd fix it soon.
Hmm.. Seems like if you Remaster with V4, kinda fixes the Muffled sound with V4.5
hello, i've noticed something: v4.5 seems to suck with remastering, but with cover it usualy does something pretty good
however the replace section is totaly unusable with v4.5. even when i completely clear the style box, the melody just goes completely wrong
Anyone know how to convert a phone number account login to a Google login? When I try the Google login, it creates me a new account, which is not what I want. I have a subscription, but no help found to change login method... HELP....
Does anyone know how to sort songs by date in individual workspaces or playlists? The order isn't chronological or follows any other obvious pattern.
Is there a way to bypass the cencor on Suno? I wantes som latin chants (for death metal) and Nigra Vis (dark power) got flagged
can i use my own voice recording and make a song ?
As far as I know, account transfer are not a thing... but it is something you can approach support and ask about, as well as provide them to the email you want to connect to.
(Many services do not allow account transfers in case someone gets scammed/phished, one reason why I am hesitate to register with a phone numer - what happens if you lose it?)
Anyways, to contsct support, see the following. Note that they appear to have a massive backlog so you will likely be waiting weeks before you get back a reply
Heads up: Discord does not have any official support team members.
For all non-financial issues, please send an email to: support@suno.com
Include:
- Your Suno @handle
- A short summary of the issue
- What you've already tried
- Any relevant screenshots or references
You'll get an automated reply when your email is received. Sending multiple emails about the same issue may reset your place in the queue.
Thanks for the feedback, makes sense. Won't make that mistake again with phone number.
Do at least try to contact them and find out. Worst case scenario would be where you are at right now while best case would be that they will link it.
What's funny is that I remember we used to have a manage account option before where you can link other SSOs to your account. Unsure why that particular page was taken down but likely was involved in some shady use by third party companies and bot users, though that is a major assumption on my part
Too bad they don't give an option to add your email address, which is linked to Google. Then there would be an option for alternate 2-part verification.
Hi I am not able to login to the web interface I am not getting any OTP from suno end to login
Now it has blocked my login saying something suspicious and blocked my account what to do
I assume you're trying to log through the phone number account? We're you connected to VPN?
Never heard of this happening before, quite unusual. Many verifiers will block attempts to log in for 24 hrs upon several incorrect inputs, though some lock it until you reach the department involved with the accounts (which would be through support@suno.com)
Again, this is if you are doing the phone number login. If you are using one of the third party SSOs (esp. Discord), this might be different but same process when a lock down happens - contact that companies support staff (usually they provide details on who you contact)
Does the account have a subscription status?
Yeah, usually that is for issues that can be handled by regular community members (or get directions provided) as there is no official support staff here.
Hmm, you may want to reach out to Mike, the Community Manager, and see if he can get in touch with a person to freeze your membership days or something similar if your ticket will take a while to get to.
Bring up the whole situation up with him first so he can provide you with some options moving forward
See #📚┃suno-school message and read the text after point 3, midway through
Although he may not be able to help you directly (different department), he should be able to tell you the current stage of your ticket or reroute the ticket to the correct personnel who deals with that problem
Thank you
You're welcome. Just remember to allow DMs from this server's setting in order to contact Mike without needing to friend him 👍
just curious is there a way to use suno without music but just to like read something. I want to do one of those talking openings before a song like a Va style before the song starts
There are things like [Spoken Words], [Announcer], and Stand-Up Comedy styles to influence that.
For the former two, I heard having quotes around the speech text helps. To improve the odds of reading out, having a longer speech gives you a higher chance (Imma test this)
Not sure if Podcast or other such styles would influence it too
ooo thank you for answering! I have never tried it either.
Sometimes the voice it gives tends to sound weird/robotic
Test 1:
Test 2:
They all generally worked out except for the last track
Test 2, Track 1 sounded quite human at the start
While Test 1 seems very radio quality
Hello guys ! My suno.com is broken : (, I open the website, it doesn't show any accounts, i can't create songs or choose model or do anything on the website, this happened when i logged in using my facebook account and it was stuck like this even after i logged out using facebook settings, deleted the suno.com's data and cookies, still stuck like that whenever i opened the website.
Thank you so much for doing this it's super helpful!
Version 4.5 is performing very poorly. I spent 500 credits testing it, and only 2 songs came out acceptable. There are many issues: robotic vocals and instruments, distortions, instruments with too much depth—or sometimes no depth at all—worse remasters, and songs starting with the chorus even though there are verse 1 and verse 2 sections that should come first, even when I strictly instructed it to start with instruments only.
Such a shame...
How do people write their lyrical instructions? Ives seen a coue different variations do they all work or is it a bit random for example (Verse - sang softly) (Verse : sang softly) etc
[] for structure, () for vocal direction. Sometimes Suno gets it right... xD
Sometimes 😅
I done my first song not knowing how to write the prompts. Can someone be kind an suggest what format I should use, (i.e. [use this] or (theses). any tips or tricks? I was impressed how my first song turned out. Thanks for your time.
[This is for music styling commands, like Verse, this won't be sung by the AI]
While --- (these are for backing vocals and will be sung)
[] is used to give instructions about the structure of the song. For example, at the beginning of your song, you can put [instrumental intro], and right below, you can add [verse].
Like this:
"
[instrumental intro]
[Verse] (Sing gently)
Touch the night sky to feel it glow
Electric spark from head to toe
Whispers slide like velvet slow
In your eyes a fever grows
[Verse 2]
Moonlight dances on our skin
Honey words pull me in
Breathing shadows close within
Feel this world begin to spin
[Chorus] (Strong vocals, add harmonies)
Your touch like fire oh it burns
Every heartbeat my soul yearns
Passion's rhythm twists and turns
In this heat our love learns
[Verse 3] (Sing with passion)
Secret glances sweet and sly
Heat of night we don’t deny
Fingers trace a line so high
In your arms I soar and fly
[Bridge]
Silent storms electrify
Stars above drawn to the sky
Softest sighs and wildest high
We ignite no need to try
[Chorus] (Strong vocals, add harmonies)
Your touch like fire oh it burns
Every heartbeat my soul yearns
Passion's rhythm twists and turns
In this heat our love learns
[end]
"
This way, you can guide the structure of your song.
When you use (), for example, (soft vocal), you’re giving direction on how the vocal will be executed, like dynamics. It can also be used with instruments, where you can add reverb, delay, etc.
you can add [drop] after the chorus, to add a drop with instrumental, try this 😄
Thanks that helps. @crude pier @stuck dust
Hello! Is there any option or prompt that allows me to create 1 minute songs? Any hint?
Thanks a lot for this explanation, it will come in handy for my next project!
- Are you using VPN?
- Have you tried using another browser to see ifnit was a browser-specific issue?
Haven't tried this myself but try to include Short in the style prompt
I know when I wanted a short jingle at the start, I had done things like [Short instrumental intro]
hey guys, something weird happened.... I got a yearly subscription from suno to my account and they took the money from my bank account. this is the second year now, transaction was successful, so far so good, but now, i just logged into my account just to recognise, there is no plan anymore... I'm back on the free plan...they removed my pro plan....wtf...
This may indicate a transaction issue.
- Verify with your bank if the transaction was successful (e.g. sufficient credit, payment processed, etc.)
- Once verified there is no issue on your end, Contact billing (type /billing)
- You may want to check with Mike to see if he has any suggestions. I doubt he will be able to do much since it is outside his department but you never know. For more info, see #📚┃suno-school message
Hey there, is there a channel to post a suggestion?
Honestly, I am just happy enough to even get the 3.x models as a free user, most services have more restriction and poorer quality models (in relation to the subscribers, e.g. mini and pro moes)
One reason why v4 is not offered to free plans are due to the bot and API users that are hijacking Suno services.
Another reason is to incentive users to pay for the service. Suno is a company, they have to pay the server and infrastructure costs somehow
There actually is in the Feedback category at the bottom of the channel list:
#💡┃suggestions-web (There are similar channels for iOS and Android app suggestions)
Thank you 😊, couldn't see it for looking
I've checked my bank account and the transaction was succesful... so i wrote a mail to support@suno.com
Err... as I mentioned in my text, you should send it to billing.
Support will redirect it for you but the difference is timing - billing would take a few business days. Support will receive your ticket in like 2-3 business weeks due to their high backload
i did that before you answered... / while you answered.... so i saw your answer after i've sent the mail...
You can always send another mail or contact Mike to see if he can redirect it for you. Not like there is much of a difference atm as the business day did not start yet for this week (06:43 AM in Eastern Daylights Time)
also, when i look for billing, there is only the FAQ thing, which doesn't help me since its not about a general question...
and you talk about mike.. who is mike? i don't know him...so, how can i contact him?
Usually companies only posts FAQs on their sites as you can see in #1294421750397796435 - this situation rarely occurs but, iirc, there is a note at the bottom of the subscription page that states to contact billing@suno.com if there are issues
See the link I had provided you in point 3 of my earlier post, the same one that first mentioned about emailing billing #📚┃suno-school message
i just wrote the same mail to billing@suno.com
Hi its there Admin technical?
yeah, i saw that, but again? who is mike? how to contact him? you write that, like everyone would know him, but I'm not that deep into "who works for suno" or so...
The link provided in the third point discusses who he is and how to go about contacting him
ah, got it.. it was the link within the linked section...my bad
There is no official support team but maybe we, the community members, can help?
I have qestion for suno
Ah, yes, it was my bad - copied the link to the message one above the one I wanted to link
If it is a question you can only ask or receive officially from the company, you will want to send your question to support@suno.com
Ok thanks
Also forgot to mention there is billing@suno.com if it is a financial-related issue (e.g. subscribed to the yearly plan instead of monthly)
If it is a situation/question that anyone can answer, you can ask here
Honestly, suno had been good to produce a 1 minute song just by having a short verse. You trying for just an instrumental? It may be quicker to try a couple and crop at a pleasant and logical point about a minute in -- is one way.
Or just put some marks and the tag [end ] on the lyric page is one attempt. Maybe "jingle, "snippet" or "memo"can help. (See samples)
I did these just now for you....
https://suno.com/s/IHS2EAgCTruF4Jfo
Heres a cropped.one at 55 sec and a natural fade one at 29 sec and 17 sec natural end which doesn't sound too bad.
See above
Thanks for the link. I'm still asleep.
ok, thanks guys, I'll see if I can make something out of it. My wife just needs short pieces for her clips. Sure, I can cut them up, but it always sounds a bit unnatural.
I got a problem with mastering, I got an error message when trying, and no matter if it's 4.0 or 4.5 ????
Just a general error or system error isn't much to go on. Most of those just have to be self-testing to check things on your end before posting in a feedback channel such as bugs 🐛 https://discord.com/channels/1069381916492562582/1082775596767117312
Waiting a bit, especially on updates, seems to resolve many of them.
How's remastering a 3.5 song work? To compare.
Yup a 3.5 when I created a new one, the Mastering works only when tracks are created on 4.0 or 4.5 seems to have failed on mastering
Anyone having issues creating stuff? v3.5 just loads for me but never generates anything, like I don't even get 2 loading tracks but only a loading circle on the button
No im not using vpn, and the problem was only on opera GX at first, i tried edge and it worked well until i logged in with my facebook account on it and i got the same issue as in opera gx
I don't know if this issue I have is a bug or if I did something that got my song deleted, but as I was making and playing back my song they started to disappear, and now their song picture and the song are gone, now I also had other people listen to my songs and like said the songs are now gone. This is what I see. My songs "Things I Ment To Say", "Running Lights", "Halfway Home".
Can someone Tell me what Happened?
That’s odd
Sometimes Suno would ignore the style prompt completely and go nuts and create a song that only focused on the lyrics. I wonder if this is some kind of cache issue? I cleared the history and cache and started over. Now it works as it should. Do you suggest that maybe start each day by clearing the cache from the browser?
Sounds like a plan. I should do that too. Good idea.
Seen other reports of this recently. Did you make all of those songs around the same time, or did any of your older songs disappear? Any idea roughly how long these songs "existed" for before they disappeared?
So I made some old songs on 3.5 that I like, but can't remaster them to 4.0, I am on free. Is there a way?
No they were new I had just finished up some final touches, thenI copied my lyrics over into Suno and had it create, I want to say the songs being created was about 30mins apart for each song creation. I don't see any of my old songs missing but I have a lot of them.
I think you have to have a membership Zeppy Free to make only allows you to do v3.5 or lower.
As far as i know you need a pro to do mastering unfortunately
Could you send me links to the missing songs shown in your screenshot, in a DM? (You should be able to access the pages and share the link from them.) Just so I can have a look at the song details and see if I can identify anything that might help the devs to figure out what happened.
Also if there are any of those that you need the MP3 file for, let me know - I may be able to get it for you.
I will try to get you the links, I know I sent a Link to my friend earlier and it di the lyric and picture video.
Oh ok thanks
I accidentally tapped it before 4.5 updated and it worked from 3.5 to 4.0
It took all my credits tho
There was a promotion ( free 4.0) to test out for users , but you had to finish your free prompt and now will need pro to do mastering or create on 4.5
Ohhh that makes sense. Thanks!
[Guitar solo] should get me the best chance at exactly that correct? I couldn't get it to work for 3 generation sets same with [Drum solo]
Hi there, it looks like I’ve hit the same (bug) issue with songs disappearing. If I manage to find them through search, they are not playable.
Even personas created from the song cannot access (play) the songs. I’ve used both the browser and iPhone app. FYI, I’m using model v4.5. Credits were deducted.
The above happens with newly created songs, if remastering an old song (ie from 3.5) the songs do not disappear (sound is oversaturated, but that’s a different issue.)
When you press Create and it generates 2 songs, do both of them eventually disappear? and for the ones that disappear are you doing anything like sharing the links to them on Discord or anywhere else during or soon after creation?
(still trying to narrow down what's causing this)
If you are able to get links to the broken songs, send me a DM with a list of all of the broken ones and I'll try to get these forwarded on to the devs (I don't work for Suno, just trying to help gather evidence of the problem.)
Also: You may be able to extend/get whole song/crop to work around this, or remaster - though these would use further credits so it's not ideal. Let me know if you want the MP3 of any of those broken songs as I may be able to get that for you.
Its there a way to remaster on 4? 4.5 generates muffled quality
Hi - Team
A week ago I sent an email to Suno (billing@suno.com) and they haven't responded for over a week. I even sent another email and they didn't respond either. Is everything okay? Is this normal?
@modest pumice Three weeks no response
Quick question. Is there a bug with Suno? Every time I try to cover an older song with a Persona. And if I add in a few extra lyrics. The Song length is always shorter than the original. And it changes up the tempo and flow in the later half of the song.
- are the ones that disappear doing anything like sharing links? No,
- have I done anything different? Not really, other than using v4.5. it I did try to publish a couple and still gone. I stubbornly tried with several songs. (So burned a few credits). Not sure if it is model related (I should try with v4.0). There is no mention of artists or songs (which usually flag them as not allowed, but the song isn’t created either). So the filter must be somewhere in the storage or client layer (although both iPhone app and browser app filter them out, or do not not allow them to be played out).
If I just cover the song without adding extra lyrics like an intro or an extra verse, Then the cover does fine. It covers the whole length and it doesn't deviate from the original. But as soon as I add in a few extra lines. The song length is shorter and the second half of the song differs from the original
I can't download my songs there is no place to tap for the download on my library
Good morning, I am working on a Duet and I put which verse each one should sing (male and female) but when I create the song, I will get it backwards (Male for female voice and vice versa) is there a way to keep it consistent?
Unfortunately, there isn't really a good and consistent way to do this.
You'd just have to roll the dice with this by keep generating until you got lucky.
Some users manage to change the verse vocal after the fact using Edit tool... so that might also be an option.
Are you using mobile app? if so, there's no option to download on mobile apps yet.
Login to the web via suno.com and download from there... on each song, 3 dots menu > Download
Are the other people who have problems extending tracks with 4.5 ? i have created manual songs and when i try to extend it it just extend 20 seconds when i go a bit back in the beat. normaly it would extend with 1 or 2 minutes extra. now with 4.5 it's not possible for me to extend a song with more minutes
I don't think it's a v4.5 thing... as it could happen in other versions too... as I've seen reports from v4.0 on the same issue a lot.
Oh i never had this issue. since 4.5 it became a thing
Me neither. But it did happen to a lot of users on v4.0, not sure about v3.5 but it's also possible.
The cover option has become ineffective due to the frequency changes in the song. It alters the frequencies so much that it sounds like a mid-frequency dip has been applied at the top end, making the track feel mono. In Pro-Q 4, you can tweak this using the Stereo Placement → Side setting. I compared version 4.5 with the cover versions, and when I A/B them while lowering the high end on the Side channel, I get the same result: the highs are reduced and the stereo width collapses, making it sound more mono.
That's something worth to report to the dev team... as no one here are able to do anything about it.
There's a channel for this kind of feedback, maybe #📝┃user-research-feedback ??
I got no access unfortunately. I work 30 years as a producer and mix engineer and i love to play with suno for creating ideas and sometimes it hits near perfection as a pre master ( I have to tweak everything in my daw to master the full track and edit parts )
How far is 'a bit' and did you try going back even further? Usually that works in that scenario , like to the beginning of the final verse? (You really do not want to extend during a silence unless you want strange things to happen).
Though I did have a song that was stuck in terminal mode even if I went one minute behind - can't recall if the issue was solved by using [break][Instant Switch to <genre>], trimming off a bit from the end and trying again, or just not bothering and made a new one...
Quick tip for fellow EDM producers: when importing your track into a DAW, use a transient layer for the kick to maintain its punch. Suno often alters the kick’s impact, but by adding a separate transient track, you can keep your kicks consistently punchy throughout the track.
Right. That channel isn't open for everyone, but they did mention in there users can use this form to submit feedback
https://sunomusic.typeform.com/to/hcFMxd2L
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
Quick question for all the Suno ninja's here - I'm trying to force songs to a longer length using v 4.5, but I cannot get any that spit out to be over 4 minutes no matter what I tell the prompt. Anyone else running into this?
This is good info... maybe consider posting it in #1252267444773326848 to share with everyone because it'd get lost in here.
Max length is 4 min.. you want longer than 4 .. you have to Extend it.
Hmm, strange, that channel #📝┃user-research-feedback message should be open to everyone. Maybe Discord is messing up the display? I once had a link saying No-Access to me but, upon clicking it, took me there... ugh, technology
Yeah I thought so too.. but when I clicked and went there.. it says I have no permission to post
Ok, and is there a way to tell the "extension" you create in the editor how long you want it to be?
Oh, yeah, they closed the v4 feedback forum because v4.0 was released (thus the model was frozen) and opted for a manual form feedback instead of a forum for 4.5 - likely so they have less paperwork to fill out themselves heh
It's used to call "V4 feedback" or something...
thats gone now i think,
Ehy everybody! Can someone help me fixing an issue I have with the site? If it's offtopic I'll delete this ofc
For 4.5, max cutoff time is actually 8 min, but it is extremely hard to get there unless you had a lot of lyrics, or instrumental tags to delay, or using a genre that takes more time than usual like classical music. I think the longest I got was a classical remixed with another genre up to 6 min
Really? oh nice... missed that!
Yeah, it is something they didn't really point out, because they prob knew the AI loves to try to wrap things up early.
Theoretically, this means that each 4.5 extend could reserve 4 min for studying the song pattern, and 4 min for the actual extension, but I didn't test out if this was the case
Sorry I mispoke. v4.5 max is 8 min.. not 4... and usually you don't have to tell it.. just have enough song structure or lyric to cover it.
i've tried adding in more verses, more instrumentals, etc and Suno keeps just cutting sections out and initial generated tracks end up right around 3 minutes. i'm looking to build a long, slow-build progressive house track that pushes beyond 4 minutes and I'm finding it almost impossible to make happen
i've modified song structure in "Lyrics" field, I've told the "Style Description" I'm looking for long form, and I've even told it how many minutes i want it to be in both fields, but I'm getting tracks that are 2:45, 3:10, etc
Yeah.. .I don't think instructions for length would ever work... they will be ignored!
The only way is on Lyrics field... but as I haven't tried to have that long form yet.. so hard to say... but the model does support that length....
And it's brand new... so it might take a few experiments....
The AI does not know what you mean when giving time (and other similar values like # BPM) because it likely was not trained for that.
It understands things like slow/mid/fast tempo and other descriptors, but AFAIK, no training for Last X Minutes scenarios
i guess the only think i haven't tried is running an extension, then running another extension on that, and then another one and see if i can push it. Just don't always love the extension "quality" when compared to earlier in the track, seems like they wash out a bit on creativity often..
To get longer tracks you need to have enough meat in the lyrics for Suno to work with.
yeah, i've got 3 verses, but it ends up skipping over one or two of the verses for some reason and forcing the song to just 3 minutes anyway...
This is the way with older models... v3.5, v4.0... so might be worth to try on v4.5 if it didn't give you the length you need.
i've got multiple chorus's, multiple instrumentals - all to force the longer duration im looking for
im using v 4.5
i even have pauses between every line in every verse to slow it down too, lol. Nothing works for some reason. Its chopping out entire verses just to fit its artificial 3 minute constraint it seems to think it has
Maybe it tend to stick to some song structure depends on genre... and even if you add more, it'd try ignore extra stuff... idk...
yeah, maybe ill switch genre's just to see if that's the case. Okay, well thought I'd check w you guys to see if anyone else had run up against this. I do appreciate the input/guidance.. 🙂
Use a negative prompt, 4 minutes, short track
That's another option but not sure what the success rate on that.
I tested i yesterday while spending 600 credits and it the succesrate for my instrumental tracks was around 78%. Most tracks ended up around 5.40 minutes. Sometimes you need to add a tag and remove one to keep the ai 'awake'. By changing tags it will go off the autopilot.
Yeah... that would change the "seed" for the AI prompt
No kidding. Ok - would you by chance be willing to share your prompt w me so i can see the structure?
Hi guys, can someone tell me why my account isn't loading? I tried to access with Discord. I can't make songs or upgrading or logging out. If I try to make a song an error pops up saying "Unauthorized". Please help me 😦
Did you try clearing your cache or trying incognito? And new browser?
Are you using VPN? Maybe there may have been a captcha that was failed? Haven't seen this before
I'll try right now, I just thought it was weird 'cause I can do absolutely nothing
I can log in via incognito, but clearly without that I still have this annoying issue
Could have been affected by some of the browser's extensions???? as it's working in Incognito mode it seems.
Clearing the cache worked. Maybe could the Discord login method be interfering with something I don't know about?
Glad it's working again with that simple method.
It's definitely annoying but yeah for webapp, sometimes it just needs a nudge to clean up.
Thanks everyone!
On Suno website there should be the MANAGE ACCOUNT section to connect your phone number and accounts such as Google, Facebook, etc
Post that in #💡┃suggestions-web ?
It is more of a statement. The manage account option had been brought back sometime recently it seems (was gone a while ago)
All my new songs just vanish for 3 days now, 2 new updates and no fix, no response from support, really frustrating!! Anybody know anything about this issue? Thx
The company staff primarily do not work on the weekend. As far as I know, the matter is being investigated but the cause of this situation is unknown.
I think there was a link to using a persona?
For support, the emails have a huge backlog. So big that apparently people have been getting responses after a few weeks from what I last heard
I'm trying to collect examples of this (I don't work for Suno but this looks very annoying and tricky to figure out so doing what I can to help...)
A few questions:
Is this happening with every song you generate?
Is there anything you do differently with the songs that disappear?
If you are able to see the songs in your recently played song list can you DM me a list of recent songs that are unplayable? (also let me know if you want the MP3 files for any of them as I may be able to get these.)
@hollow hare thanks, I’m not using persona so it must be going beyond that
Are you also describing these features in the enhanced styles window>? The engine pays as much attention now to that or maybe even more. I had a real hard time getting Suno to give me the structures I needed and when I put them in the styles and then called them in the lyrics it finally did what I wanted...
So now chatGPT confused me. I use Suno since months now and I always write my prompts in "style", and obviously lyrics in "lyrics". And today I asked ChatGPT to help me with a prompt and it wrote me 200 words and I asked him if I should write that in "style" and he told no in "prompts" Wtf...There is no "Prompt" place right? It's in "style" right?
The video embeds for v4.5 songs are degrading the audio. Ngl this is a major disappointment considering older versions didn't suffer from this. Suno is ALL about the music so to sacrifice any bit of quality when sharing kinda defeats the purpose.
They are Meta Tags found in [brackets].
Yeah I know meta tags that I use in 'lyrics' such as [chorus], [interlude] etc... But Chatgpt confused me talking about a named "prompt" place
It told me not in "lyrics" not in "style" but in "prompt" but there is no such place
While ChatGPT could help on the prompt... it probably doesn't know a lot of things about Suno - just do what you used to do.
Yes you can (and have been able to) affect the track within the meta tags, and in 4.5, its even more important to combine them with enhanced styles..
Yeah I think it is this
Thanks guys, you de-confused me lol
i write a fair bit of Irish music and Suno would not follow traditional song structures until I started describing it in the enhanced styles and then calling them out in meta tags. It was a day long battle figuring that out.
I've started using the enhance styles too today, it's for why I got confused. But what do you mean, call them out in meta tags? You write some of them among the lyrics for Suno to follow indications?
@unique stone it happens every time, I’m hitting “re use style” and making some edits on a song. It generates and lets me listen to it once, then it vanishes from my library when refreshed. If I hit “search library” I can see it blurred out and it won’t play again, as if I deleted it. I did a test creation and that worked fine so far, so it seems to be when trying to reuse and edit. I don’t know how to get links and send them but I’ll try to get a screenshot
Yes, you would designate how instruments play in the meta tags for each section. It can get very complex.
On the home page if you scroll down there's a "recently played" section - it should be in there. Typically where I've seen this happen you can open the song page from there and then copy the address of the page - or press the share button (should be on the right)
If you can do a screen recording, that would be ideal (so, record before you do anything, and then stop the recording when the song has disappeared).
There may be something useful in that which might help pinpoint the cause.
Nice, I should do that more, cause until them, I keep it simple
At least when something's important for me
Ive been using Suno for 6 months, been through 3 different models and each one has been a learning curve 😛
Yeah, same for me...I'm still learning
Grok seems to understand Suno a bit better than GPT, you might try experiment with prompting direction there...
Is it this option you were using?
S/O to everyone helping other people here! ALWAYS a big deal to help people! Thank you for what you do!
@unique stone reuse style, iOS app
Ahh ok, interesting... the previous reports have been of desktop browser so that rules it out as being a specific browser issue.
@unique stone I sincerely appreciate your help, thank you. It’s probably just a waiting game at this point, sounds like they’ve got wind of the issues at least
So i put male vocals as one of the styles, and i keep getting a female voice. Any idea on how to get that to be a male voice?
Male vocalist what i use..
I’ve been using version 4.5 to generate songs, but I’m running into issues. The vocals have a robotic quality, there’s noticeable hissing, and a lot of background noise. I tried remastering, but it actually makes things worse. Could you please look into this or release a patch? It’s affecting all of my songs.
Could you share some examples of prompts or techniques to get clearer sound quality?
Share your feedback with the team: https://sunomusic.typeform.com/to/hcFMxd2L
Turn data collection into an experience with Typeform. Create beautiful online forms, surveys, quizzes, and so much more. Try it for FREE.
Please link suno songs as example. uploaded wavs are not usable for diagnosis.
Here is the link to a song with poor quality: https://suno.com/s/SkOQ1om2t71Igxhd
It sounds pretty good to me, to be honest! but maybe I don't have the best ears!
check this one : https://suno.com/s/T0ggrO7DdbQ2RHND
@unique stone I tested with another brand new creation and it disappeared so it’s not only with the reuse style function 🙁
Can you send me a link to this one?
Thanks for the update
I like the song! Don't understand the words, but I like it and can't hear the problems!
@unique stone it’s not in my library but it was in the recent played on Home Screen. It also didn’t blur out yet but they either blur out right away or after a few minutes, so it probably will blur out soon and be unplayable
Do you get to actually play these before they stop working? Any idea roughly how long after creating them they stop working?
@unique stone yeah, it plays one time, then when I go to my library they wisp away. If I hit search library it shows there but usually blurred out immediately and unplayable. Sometimes they are still playable for about 5-30 minutes then just blur out
hi, can anyone help, my songs are self deleting if that makes sense, if i refresh the page they're gone, i've wasted at least 200 credits. v4.5 btw, some were covers, but all songs are disappearing
@unique stone the app is having other problems too, some icons are non responsive. When full screen cover art is open, it won’t close back down to see library or playlist, can’t switch songs without toggling all the way through every song or closing the app and opening again
Same here, lost two good tracks Saturday evening and emailed support yesterday. Waiting for answer, hope I can get ehm back somehow
@sour laurel I’m having exact same issue for 3 days now
is this a known problem?
For app issues I'd recommend https://discord.com/channels/1069381916492562582/1258072243963957288 - I don't have the iOS app so I'm not the best person to help with that one 😅
oh so it is
Kind of, haven't heard anything from Suno about them working on it or being aware
same here, messaged them too, now i just download everything. is it only v4.5 for you guys or v4 and 3.5 as well?
Just heard a couple more having the problem
Only work with 4,5 since it came out so hard to say for me
No answer yet?
my v4 songs seem to work just fine, just checked
nope, messaged them this morning, so 12 hours ago give or take
hopefully
why is the audio upload limited to 2min?
@median crest @sour laurel
If you can find the songs in your "recently played" list and can access the pages for them at least, can you DM me with a list of the ones that don't play? (Just so I can have a look in case there's anything interesting or unusual about the song info, and forward it on as further evidence of the problem)
Also if there are any that you want the MP3 file of, let me know and I may be able to get that for you.
(Not Suno staff - just trying to help gather info that may be useful for the devs)
Sure, thanks!
I hate to say it but 4.0 sounds less AI then 4.5
V4 is working fine for me, so it must be v4.5 with the issues
To the Admin and Developers, tested ver 4.5 , for Tamil Language dont retire Ver 3.5 , Posting the same in Language chat too.
Hi,
I’ve got the feeling that persona only puts the same « style ». Cause I always have the same Melody…
Is it possible to use a persona and change the style to have another melody ?
Yes, it is possible to have a different melody. That said, they do have a strong bias to recreate the melody of their original song whenever they can. It might be possible to change this if the lyrics you use cannot be made to fit the melody, but I can't say for certain
Help: My persona only sings the same melody now no matter the genre or style. How do I use the persona on a new song?
Hello, I bought the $96 subscription but I wanted to get the $10 one. I thought there was an $8 offer and I rushed to get it. I can't afford it, I sent an email but I haven't received any response. Please help me.
Like does it ban you ?
What's this new error saying actually?
An error occurred.
Invalid request. Need all required params: artist_clip_id, continue_clip_id.
Extend + Adding a Persona
Could it be because the persona is not v4.5?
Looks like Persona's are not being able to be made for me, on v4.5 not sure if I'm the only one?
i have the same error for "cover"
Ah ok, so it's a bug then
@wet minnow Heard anything about this? Do you mind trying?
i have a feeling that might be connected to the "breaking songs" bug... there are details missing on the song pages and the songs dont play. @unique stone
Anyone k ow why I am getting this error An error occurred.
Invalid request. Need all required params: artist_clip_id.
oh oh
i have the feeling we are on to something here
An error occurred.
Invalid request. Need all required params: artist_clip_id, continue_clip_id.
Hello guys