#🪲┃bugs-web
1 messages · Page 7 of 1
When this is happening it's an entirely new generation even though it's a continuation
I didn't see that on the other clip because that one wasn't the full 2:00 minutes. This new one, however, is, and should only be 3, not 4 minutes
It does not seem to matter if it has lyrics or if the Instrumental toggle is on.
I'm having the same issue @viscid garden has described. Only started tonight alongside the limited generations banner appearing.
Showing how it looks to me. That's the part 2 that transitions at the 2:00 mark in the song, but the whole song is 4 minutes rather than 3 minutes. Because they are two unrelated new generations mashed together
I'll take a look in a bit. I'm not sure if the problem is on the UI side (the form didn't continue from where you set it, instead used end of clip) or on the server.
@neat sail In my case, when I'm getting my 2 continuation generations, 1 of them will typically be correct and the other one will randomly maybe be a "fresh start". But both are labeled in the usual way.
That could just be AI quirkiness. Like even the sound of a "record needle falling" in the previous clip makes the AI think "Oh, NEW SONG, that means NEW SONG!"
Maybe, but it has never done that before tonight, so it's definitely a new behavior.
It's happening often enough that I'm holding off on new generations for now, because it ends up wasting a lot of my time and credits.
Is it happening often on brand new clips you continue, or are you trying to continue from the same clip?
I think it's happening mostly when generating Part 2's.
I've been Trash Canning the ones where it happens, but I can probably dig some up if it'll help.
I notice that lyrics regularly disappear from the song. Then in the continuations, there is regularly also a lack of coherence with the previous part (whether instrumental or vocal)
for several times in a row, only the second generation of the two continuations is actually a continuation, another one being a from-scratch generation. Flagged as a bug
Did you check if your payment? Maybe it went wrong there?
I'm not sure exactly if it's 30 days, or exactly 1 per month, but it's probably just not quite at the reset time yet
For billing issues you'll have to email billing@suno.ai though, moderators can't do anything about that type of problem
anyone experiencing any difficulties downloading audio? happens nothing when I click. the song is 9 minutes long so that might be it, but it never happened before
i'm trying lotsa guitar based styles like Shoegaze, Darkwave, Post-Punk and there is this tendency of the results DROWNING IN BASS.
the bass even gets stronger towards end of the song and distorts everything else:
Hello, I have several creations on the site, however I noticed that some have disappeared, am I limited to 20 songs at a time in my creation list?
Click the library menu option. The create screen only shows the most recent.
go to Library
oh, you beat me, hehe...
Thank, you save my heart ❤️
here's a mastered version of one of the tracks from about 6 posts above where the progressing bass distortion can be heard better.
i had to take the bass back -12 in Audacity and an additional -2 for the second half. starts out quite ok, but then... 😢 :
yeah, well... i tried.
Not sure if this is an actual bug. Someone may have already mentioned this. V3a has a habbit of hallucinating. In some cases totally forgetting what it continued from. From different vibes to changing the singer from one gender to another and again changing the whole track mid song lol. Just thought I would mention it just in case.
Yeah the said it would halucinate etc. See the announcement here: #📢┃announcements message
But alpha jkjk? But keep sending in bugs!
Well stop giving the poor A.I. shrooms! BWAHAHAHAHA! Couldn't help it.
Songs cutting off very often. I rarely get the full 2 minutes with v3 or 1:20 with v2.
Also depends how long your lyrics are., in combination with the bpm of your song. For instance a ballad is slow, so then you don't need much text in the lyrics box.. However if you have rap, you don't have enough text to get to the end of 2 minutes.. So it will chop it off before 2 minutes then if that makes sense.
is there some tag or something to use to not get completely ignored here?
i posted examples and very clear descriptions.
even a "i don't know" or "eff off"would be better than just crickets. 😉
Another thing I have noticed: Anything in [] it will still sing sometimes. Again probably already mentioned lol. However, what I put in the [] probably may be the reason. Something like this:
[Outro]
As the final notes fade, and the spotlight descends,
A choice faintly glimmers, a tale with no end.
In the wreckage of sorrow, where demons convene,
A glimmer of hope, a path yet unseen.
[Instrumental Outro - hints of melancholy with a rising note of possibility]
v3 does that a lot, unfortunately. i usually only use the most basic one word tags like [intro], [verse], [chorus], etc with v3 with a single more descriptive one-liner at the start and maybe end. like your last line you posted above.
I noticed that putting [Break] before the lyrics when continuing a song it helps in v3
Not sure how many have mentioned this. But just in case. There is the message that I have only seen once or twice while using v2 or v3 about this is copyrighted etc etc. When it is a repeated chorus. Unless it is the progression or continue from or something. I almost promise it isn't copyright material lol. And I don't want to change anything. I changed the chorus slightly to see if that helped. No dice. Which is why I think it might be the sound or music itself.
Maybe the title? I'm not sure if that might be a bug where the title is being checked as well, but I wrote it down to check later, just in case.
will change the title. However this song is almost done. This is the only time it did this throught all gens\continue from.
didn't work. i even left the title blank. but it is probably ready what was continued as well. man oh man i dont want to lose this lovely swamp jazz solo. The issue came from trying to continue from this: https://app.suno.ai/song/8a79d45d-21a6-4d84-aa95-e19c96d66f80
Dark Cabaret, Electronic, Electro Swing, New Orleans, Swamp Jazz song. Listen and make your own with Suno.
reading not ready lmao
Hi there, today for the whole day my continuations were on and off. Many times it gave me a continuation for a different song or it was a completely different voice. Just mention it, maybe it is already mentioned but maybe it need some investigation
If you happen to have a really clear example of a continue clip problem could you DM me the relevant clips. Ideally both clip 1 and clip 2, if clip 2 is the continue. But just the continued clip is better than nothing if you lost track. I haven't dug into the reports yet but I have gotten a few similar reported bugs and plan on digging in a bit, sometime later this week.
Also just in general, if anyone thinks they have a clear-cut bugged continue I would appreciate DMing me the relevant clips. I'm particularly looking for something that couldn't be attributable to stubborn or quirky AI, like the continue time being just plain wrong.
From this: https://app.suno.ai/song/a0355195-e691-4b74-8e56-4a08cd28551c
To this: https://app.suno.ai/song/d33964dd-7acc-4381-823a-f26fc0363816
Downtempo synthwave song. Listen and make your own with Suno.
I will if and when it happens again. I can't remember which of my clips did this but several did. I will share the clips when it happens again.
Also, for some reason the last mp3 generated was rejected by soundcloud. i tried others generated in v2 and v3a. I added the generated mp3 file just in case and also the Suno link. Soundcloud says it's not valid. Also when i converted it to wave file same thing. I checked the details, i cant find any reason why soundcloud would output that error. https://app.suno.ai/song/72feab4c-8829-4768-a37c-166d1264e1e8
I have downloaded it twice with the same issue.
Dark Cabaret, Electronic, Electro Swing, New Orleans, Swamp Jazz song. Listen and make your own with Suno.
Nevermind, i refreshed suno went to the profile page and downloaded from there and all is good. Wondering what happened in the first place tho.
fix this! XD
your mp3 file is perfectly uploadable on soundcloud. 😉
Why i can't change model to v3? I don't have that option
it is now. maybe i didnt wait long enough or something. the first time i downloaded the mp3 soundcloud sent an error. It could have been a temp issue on SC side?
not sure what exactly happened. but what i know is that SC checks the file for copyrights,
which should only apply when uploading an original track by somebody or a version of it.
having it in a dj mix seems to be ok on the other hand.
that wasnt the issue. it was saying the file was invalid. but i believe it was a soundcloud thing not suno mp3 now. working fine now.
Seems that V3 can't produce more than 18 seconds for a Part 2 but is happy to take the credits. Wasted almost 1000 credits down to 0. I kept saying to myself to stop paying them to test their own version but I had one not too bad start and tried to finish it as it can't go from V3 to V2 for that. The very last 5 credit render was longer than 18 seconds but nothing like what it was continuing from. But it took the credit. Oh and I think this is a BUG, most vocals now sound like Country Music even with a Dance House style.
maybe you tried an mp4 ? 🙂 i know that this absolutely would produce the INVALID message.
There's a major problem with percussion. Most of the notes are pure static and tend to ruin songs.
Since the last maintainence I've had multiple incidents of the second part of a song I generated creating a two minute clip of a new song, instead of extending the current track for 1 minute. Here's one example https://app.suno.ai/song/65697376-cda0-4bbf-bc80-94f7a06703a1
Okay I can personally confirm this double length continue bug now. Has anyone had it happen to them going V3 -> V3? In my case it was V2 -> V3.
Hey, I have Noticed in my latest generations that whenever I continue a song the second version of it is completely different from the previous part
This happens with version 3.
Was the clip that was very different also extra super long?
This one would go wrong several times, I have the bill paid and this thing twice doesn't follow the style of music. You ask for rock and he puts in country music. you ask for pop and he puts out something completely random. other than that he often doesn't put what we write in the lyrics., there are days when he is unbearable, other than that he often doesn't finish the song leaving it practically infinite because she doesn't want to finish the song, there are days when he obeys correctly but there are days that just makes me angry
In v3, I’m getting a pretty regular repeating of the chorus multiple times. This very rarely happened in v2.
Feels like approximately 25% of the time. https://cdn1.suno.ai/5a2eaa17-7194-4013-b4f3-024cd1fe6730.mp4
In above example, after Outro, Chorus and Outro repeat
This is still happening, btw. It just happened again on a first continuation of a track. One continuation was perfect, the other one was totally unrelated to the source song. It doesn't pick up any of the structural cues at all.
did the clip you continued from end with silence or something like that?
No. When I continue a clip, about half the time one of the two continuations appears to be a completely new song, unrelated to the one I'm continuing. This only started happening yesterday.
There's absolutely nothing unusual about the clips I'm continuing. I've done literally hundreds and this never happened, now it's happening about half the time.
is that all on v3 alpha, or is v2 also doing it?
I haven't used v2 since v3 alpha came out, so not sure.
I can say v3 alpha was not doing this until yesterday.
Sounds like a good clue. I haven't seen that behavior yet. I also haven't been using it much today.
It's driving me crazy, and wasting a lot of my time and credits.
excuse me while I waste credits trying to get the same problem, lol
😲 LOL
I mean, in all seriousness, I am not even close to running out of credits (try as I might). It's just the principle of it, plus the wasted time.
https://app.suno.ai/song/3322c7dd-c6b8-4f75-adaf-aec8381763d4 when continued gave poor man's canon in d major: https://app.suno.ai/song/5c8ea617-b825-4449-b33e-07c3bd81ebed
Instrumental Interlude, Medieval festival song. Listen and make your own with Suno.
Medieval festival song. Listen and make your own with Suno.
Just once though.
I can't get a v3 continuation either, however switching back to v2 seemed make something that makes sense at least sometimes.
When I try to continue the track, for some reason it changes the language, which makes the text lose all meaning. Is there any way to fix this or specify a specific language in the request?
im generating the same songs only a slight difference in instruments and melodies, but the tempo, key, and structure of the song is exactly the same
this is also happening to me, i even have continuations that are breaking into three songs
Since this afternoon, songs written in Chinese have always come out with works inconsistent with the lyrics, and sometimes they are not sung in Chinese, which makes me very headache. What is the situation? Is it better to make English songs?
Me too
so do i
so do i
Since this afternoon, songs written in Chinese have always come out with works inconsistent with the lyrics, and sometimes they are not sung in Chinese, which makes me very headache. What is the situation? Is it better to make English songs?,I wasted 500 points.
Since this afternoon, songs written in Chinese have always come out with works inconsistent with the lyrics, and sometimes they are not sung in Chinese, which makes me very headache. What is the situation? Is it better to make English songs?,I wasted 500 points,Please reply as soon as possible and solve this problem. I don't want to waste time, energy and money. Thank you.
Sorry to hear about the issues with Chinese songs. Suno's continually working on improving language support, including for Chinese. It's always a work in progress. For now, you might find more consistent results with English songs.
We make announcements if we do any quality updates on the models. Otherwise, the errors you run into is likely caused by just luck or bad inputs. I'd recommend change your lyrics or how you prompt it.
sometimes songs are generated without vocals although the words are in the box
yep it happens...should be much less often in v3 than in v2
The cover art generator just created an image of Sonic the Hedgehog. Reported as inappropriate in the app
There's something fundamentally wrong with V3. After using it for days, and wasting tons of credits, it's clear that when you continue a perfectly good first part, the next part almost always mangles the lyrics (assuming you get past the prompt ban!), by missing words, missing whole lines, jumping back to the first verse, repeating lines, singing the previously great chorus in a totally different way, it's a mess.
The issue you're facing with V3 of Suno, where the AI continues from the previous verse instead of using new lyrics, has been acknowledged in the discussions. According to a response from a Suno team member, this happens because the model thinks it hasn't finished the previous verse yet. To mitigate this, they suggest continuing from a point in the song where the previous verse didn't start. This could potentially prevent the AI from reverting back to or repeating the previous verse.
serious question: what do i have to do to not get COMPLETELY ignored here ? it's so extreme that it feels like i'm shadow banned or something, and i did nothing wrong.
i'm wasting away credits for songs where the bass continually gets louder in every continuation, distorting more than half of the song so much it's basically destroing it.
The issue you're experiencing with the bass getting louder and distorting the song in each continuation is a recognized problem. This distortion occurs because the AI tends to amplify the noise present in the initial clip, which then increases with each continuation. It's a known challenge with the AI's creation process. The AI hears noise in the first clip and makes more of it in subsequent generations. This phenomenon is being investigated and is a focus of ongoing research to improve the system.
Furthermore I like to explain to you that I am a volunteer guide. I am helping people to the best of my ability. I am not a Suno Staff Member 😉
😻 this is the sweetest and most concise response to a technical question i ever got ! also the only one... 🔧 ☮️
No, I carefully continue the words correctly, this didn't happen in V2.
Despite your careful continuation efforts, it seems this issue persists in V3 and differs from your experience with V2. This feedback is valuable for the ongoing development and improvement of the V3 model. ❤️
I love your reply and a lot of songs fit this category. However, there can be exceptions: I generated a song some days ago where it gets better with every verse. The first verse is the worst, full of background noise. The last part is the best. The AI is so unpredictable 😊
🙂 i think that's why it is called alpha version @craggy charm . Alphas are highly experimental. I'm so glad the suno team released the alpha version and gave us the possibilitie to burn credits on the alpha, because it creates wonderful pieces ❤️
NGL, it is kind of an honor for me to be able to use suno ... as a software dev I'm amazed
created this prompt after using continue from a previous one where i only chose to make a instrumental version, but on this prompt i chose to add lyrics from chat gpt. I think it still just made an instrumental version and lyrics were not used.
tried the same in chirp-beta-7, seems like it does the same thing for the first version, however it was correct on the second one.
that's either the most subtle sarcasm ever, or you're just a sweet person. 😄 i wouldn't mind if alpha were free, since in this case the testing and feedback by users could be considered currency. 😉
I asked Suno lyrics generator for a generic high school sports team pep rally song. It produced a song that contained the generic term "We are the champions", which then generated a copyright error.
high school marching band, cheerleaders, trumpet, cymbals song. Listen and make your own with Suno.
Don't get me wrong, I am getting some really good tunes in V3, particularly instrumentals, I have some blues and jazz fusion that is blowing my mind.
Yes, I understand that. I meant something else - the song is not even near the ending and it cuts off like in 20-30 seconds, even when there are lyrics left. Sometimes it fixes itself after many times of re-generating, but it's wasting many credits. Refund option of credits would be perfect for these types of scenarios.
This could stem from a technical glitch that occasionally resolves itself after multiple regeneration attempts. Offering a refund option for credits in such cases would be advantageous for users experiencing these issues, helping to offset credit losses from unsuccessful song generations. The Suno team is currently evaluating the best approach for refunding credits, aiming to strike a fair balance that prioritizes user satisfaction while upholding the integrity of the credit system. @velvet ridge
I Keep getting spammed saying its failing for copyright its upsetting its my own lyrics
It is some stupid copyright detection bot messing up, sorry to hear that. The Suno Team expect to have a major update on this soon.
OK
Seems like v3 is available on basic plan after your subscription ends. Is this intentional?
Additionally I seem to be able to use my top-up credits without an active subscription (these are the free ones given to subscribers from when v3 alpha was released) which the subscription page suggests is not possible ("Purchased top up credits do not expire, but require an active subscription to use.") Are the free credits we were given recently different to the purchased topup ones somehow?
In my opinion the Alpha version can be considered free, given the extensive resources and services provided by Suno, including GPUs, DevOps, networking, processing, and maintenance behind a functioning interface. This setup involves a team of developers who are compensated monthly. When considering the costs of credits or the credit-cost effect, the provision of such comprehensive infrastructure and human labor for the development and maintenance of this Alpha version suggests that bonding and playing with it are essentially offered at no direct cost to the user.
Why?
The Alpha of the V3 is now way better to generate Hardstyle songs. However, it seems it struggles with the metatags. For exemple, if I want a [Drop] or [Instrumental], there will be only the melody for few seconds then continue on my lyrics, without kicks. I have to generate a lot of chirps in order to have something that matches an usual Hardstyle structure song.
V3 generate 2:00 song with nothing in it. I got that multiple times.
Major issue:
Says that prompt is copyrighted. False, very damn false. I NEVER take lyrics from AI or Google. I try to make stuff myself.
So hello???
Suno's system sometimes generates generic error messages when it encounters issues with prompts, such as when it identifies something as potentially copyrighted, offensive, or too closely matching existing lyrics or songs. This can be confusing, especially when you are certain that their content is original. Suno acknowledges this issue and seems to be aware of the need for more specific error messages. However, the development of such specific error messaging is still in progress.
I understand. Thanks for your answer. But it's so frustrating because I KNOW they are MY lyrics. Not generated, not copied.
Trust me, I feel your pain. I have the same experience with some of my lyrics! For now, if you're sure your lyrics are original, you might consider slightly altering your prompt. Keep in mind that the system might incorrectly flag your content, so it's a matter of trial and error until Suno can provide more precise feedback mechanisms.
Aighty. Thanks so much for your assistance here. Actually you are right, tried modifying slightly and it worked out.
Cheers!
So happy to hear that... Have a good one!
Hi guys not sure if it’s a bug but currently extend songs only goes for 6 seconds?
I'm sorry to hear that you're encountering an issue with song extension duration. In Suno, the extension duration of a song should typically be longer than 6 seconds, depending on your specific settings and the song's context. I recommend checking if the song is already fade out or has silence on the point you want to continuate from. If that is the case, I would suggest to start the continue a tad earlier by using the new feature to start from anywhere.
Hi! I am a Pro member, but I can't find any way to switch from V2 to V3? I see no toggle under /create, and all songs generated are marked V2. Thanks
There has been an issue reported with V3 Alpha not being available for some users. This is often related to billing issues. If you don't see the option to toggle to V3 Alpha, it's recommended to check your account page and ensure all invoices are paid. If everything looks fine on your account and the issue persists, it's advised to send an email to Suno support for further assistance. You can check your payments via "Subscriptions", then "Edit Billing Details" on your account page.
thanks!
when I try to Remix but off Custom Mode;then while I click Create, Suno Even though My Description is in Japanese and sure not related to any artist, Suno still said that:"Couldn't generate that.
Song Description contained artist name: tobu" in English, has anyone know how to fix that?
My text in Song Description:
"
瞳に映るで極彩色
それらをすべて一緒に注ぐと、カオスと濁りだけが表示されますが、数百、数千の顔料が混合された濁った黒色でさえ、比較するとはるかに鮮明です。
混乱と不確実性の真っ只中には、怒り、悲しみ、不満が満ちています空にむせ返るような塵を伴います。
人々はその中に閉じ込められている,飛び散った絵の具で自分を鮮やかに描こうとすればするほど、絵の具に染まってしまい、澄んだ色を失ってしまう。
"
I have no idea that where would be consider related to artist name: tobu, but seems AI somehow consider it's related.
(Pro Plan) Very often (1 on 6 times) on V2 or V3, there are no lyrics in the generation (despite Instrumental only is NOT selected), can you detect this and give my credits back 😉
"kalimba" gets flagged as an artist for some reason
Be nice if there was a gap at the start of the vocals, All the ones I've done on V3 the first word bangs in right at the Beginning, often loosing its pronunciation
Tobu is an EDM artist, iirc
Thank. By the way, I now found which words in my text leading to that:い、澄, this three words combine together would lead to sunoAI consider describing contain Tobu😓 It's still very strange...
Yeah, indeed it is
I haven't used suno in a while, waiting for the model to go back to initial levels of fidelity
Yes, copyright is a big pain. The issue with the "Couldn't generate that. Prompt likely copyrighted" error is known to give "false postives" for some, and Suno is actively working on fixing it. It can sometimes be a bug and not necessarily because the lyrics are copyrighted. If you face this issue with your own lyrics, it's recommended to change the lyrics or how you prompt it and try again.
You can try doing that without overly censoring yourself too.
Assuming that it's a genuine glitch (which it sounds like it could be) try adding several lines of indeterminate length of spaces at the end of your lyrics. Add some adlibs (oh--, yeah!, etc) at various points as appropriate. Add metatags to the end for instrumentals or just make up tags, its about changing the profile of the song so that the AI will see it as something different than whatever it is that it is blocking.
I continued from the middle of a part, cutting some verses from the end and in the next part the bot hallucinated verses from the ones that were cut and should be unknown for the next generation. This might be a bug to investigate.
Thank you for letting the Devs know!
It seems like the issue you encountered, where the bot hallucinates verses from parts that were cut, is recognized as a potential problem when continuing from a certain point in a song. When the bot sings random words from elsewhere in the song, repeats a section from the last clip, or sings gibberish hallucinations, it indicates that it has lost its place in the lyrics. This can happen for various reasons, and there isn't a one-size-fits-all solution.
The best way to continue lyrics in the next clip depends on the pace of the song. For slower lyrics, try picking up with the next word. For faster lyrics, adding a line or two that have already been sung as a lead-in can be effective. If these methods don't work, you may develop a better sense of which method works best for your particular song through trial and error.
If the lyrics go completely off track on both clips, the issue might stem from the clip you're continuing from. In such cases, it's advisable to go back to an earlier clip and regenerate the clip you're continuing from, hoping for a better result.
So I'm working a lot with V3 but I think it has problems with sounding like autotune too much.
Having a lot of issues with continuations suffering from PE (Premature Endulation)... the clip just suddenly and abruptly terminates 20-25 seconds into the lyrics when it should have come out around 55 seconds. This is different from the issue where the engine decides that a song is ending and switches to a fade out or other closing phrase like repeating the chorus. This is an instant and abrupt stop in the middle of a word or line of lyrics.
I understand your concern about the V3 version of Suno's AI sounding too much like autotune. This issue can arise from the way the AI interprets and processes the music generation prompts. Since V3 is still in its alpha phase, it's not uncommon for certain aspects, like vocal processing, to have room for improvement. The Suno team is likely aware of these issues and is working on balancing and refining the technology for better outcomes. Your feedback as a user is valuable for this process. In the meantime, you might want to experiment with different styles, genres, or prompt structures to see if you can achieve a more natural sound. Additionally, keeping an eye on updates and participating in community feedback can help you stay informed about any improvements or new features that address this issue.
Don't get me wrong, Suno is some dope shit! I love it but this is one of the few bugs I noticed. Together with some rare cases of hallucinations
The issue of premature endings in song continuations, resulting in clips abruptly stopping after 20-25 seconds, could be due to the original clip (from which the continuation is generated) already fading out or having silence at the point of continuation. To address this, it is recommended to use the "continue from" option and choose a different time to continue, ideally a few seconds before the silence or fade out starts in the original clip. This approach can help avoid the abrupt stop and ensure a more seamless continuation of the song.
So I found what is clearly a 'bug'. In my song, vocal output DEGENERATES (or tends to degenerate) as the song grows longer: The opening verse starts out EXTREMELY clear in the first two or three iterations, but as I get closer to the end, it updates the beginning of the song, losing TONS of clarity. This is a big problem. Thanks.
I understand how frustrating it must be to encounter a decline in vocal clarity as your song progresses in Suno. This issue might be due to the AI's handling of longer compositions. AI models can struggle with maintaining consistency and quality over extended sequences. This can result in a degradation of clarity, especially in complex or lengthy tracks. As a workaround, consider creating shorter segments and then combining them, or using the "Continue from this clip" feature to extend the song while maintaining quality.
Sorry, I should have said that indeed I DO use VERY short clips (4 line verse/chorus) and I DO use 'continue from this clip'...The problem isn't a 'degraded song' per se...it's that the ORIGINAL OUTPUT degrades as I continue to iterate as a function of 'continue from this clip': In other words, a GREAT original beginning when combined via the 'continue' functions slowly degrades over time with each lengthening process...What begins as very clear words slowly turns into 'mush', where words are slurred like the AI is 'drunk' and the sound is mushy and unclear...This is a BUG and a problem. Thanks for answering.
I understand your concern about the degradation of vocal clarity in your song as it progresses, especially when using the 'continue from this clip' feature. Your feedback is invaluable in helping to identify and address such issues. Thank you for reporting!
i didn't actually complain 🙂
This is why I don't love people do very short clips to piece them together -- you don't have the perfect context in some sense. If there is a tiny "noise" in the first clips, the later clips will likely to escalate it.
I'd suggest put in the same genre prompt in continues, if you haven't (unless you want to have different styles in the middle).
the new length of v3 is pretty sweet. now all we need is its acceptance of EVERY [bracketed prompt] as a prompt and not lyrics. like v2 🥹
yeah...I know
Is there any way to dispute a content moderation flag. My lyrics are getting flagged and they shouldn’t be. Particularly since I am paying.
I understand the issue with your lyrics being incorrectly flagged by Suno's moderation system. The Suno team is aware of such challenges and is working on improvements.
In the meantime you can try to make a workaround by changing (part) of you lyrics. I had to do it a few times too 😅
p.s. in no way shape or form is it related if you are a paying costumer btw.
Hello, I would like to report a problem that I just encountered on the platform. I am a Pro Client and currently depend on the platform to create my projects. Next to each song, there are three dots that represent ellipses and should give you the option to extend the song to create a complete version. However, when clicking on these ellipses, no option appears to perform this procedure. Could you investigate this bug?
Thank you for reporting this. The Team has been alerted about this issue and will be working on a resolution soon.
Thank you very much!
is it just me, or does the ellipsis button no longer work?
ok, not the only one then
that's a drag, wanted to spend the evening composing 🙂 😦 🙂
oh, it works on mobile
here's a hack: make your window narrow so that the mobile version appears, it works in it
unable to download song. Nothing happens when I clicked on it, facing this issue since today morning
look straight above you
Yeah, read the previous msg
oh my bad.
Really why should they fix the bug, let's look for a way to make it so that they do not change anything, what a stupidity!
I forgot how fun discord communities can be
That's not stupidity... just a work around so we could continue with our song and not blocked waiting for the fix
its called a workaround... they fix it and we can keep using it with this hack during the fixing
@dull summit @lusty pollen @full socket @wind ginkgo I know it is frustrating. But the workaround @unreal garnet came up with, is at least something right?
You speak as though they're not going to fix this as soon as they can. Of course they're going to fix it. This is just something that lets someone in the middle of their song keep working so that they don't lose the inspiration.
I'm not complaining
as long as it works for me
should be fixed now, thanks for catching!
wow you guys are fast
Hi, the 3... dont do anything right now. It this correct or a local problem?
who needs workarounds when you can complain
Refresh?
chrome, firefox, edge working
And it's SATERDAY.. boys and girls..
check convo a few post up
@hot yacht: i even tried a different browser (from firefox to edge)
refresh the page
It's fixed!
Must be local cache... my works with a simple refresh
Yeah maybe... because it works again 😄
@hot yacht: You were right! many thanks. deleting all brwoser data solved the problem. many thanks again!
btw... are there any promo codes out there? 🙂 ok, that isn't a bug
Not that I am aware off atm.. I would love them too.. paying for 2 premiums 😉 😉
I am aware of that and it should be. Just saying if your paying then you should be able to use explicit lyrics. Its my understanding that the paying customers get commercial rights. Then, if thats the case the lyrics that are mine owned by me shouldnt be censored or moderated. In this case I wanted my intro to be the riflemans creed from the marine corps. So I get the argument I just disagree. And the lyrics I was using in particular shouldnt be moderated but, I still love SUNO its amazing and I am sure some of this stuff will work itself out. It hasnt been that long and it has gotten supremely better in a very short time. So not complaining just thinking out loud I guess.
I understand your perspective, Manimal, and it's great to hear that you appreciate Suno's rapid improvement. Regarding the use of explicit lyrics, even for paying customers, Suno adheres to certain content guidelines to ensure a respectful and safe community for all users. While you do have commercial rights over your creations, these guidelines might still apply to the content generated through Suno. I see why this situation can be frustrating, especially when specific content like the riflemans creed has significance for you. I will make sure your concerns are heard, that's the least I can do as a volunteer guide, but I will do it gladly since you do have a fair point.
Fair enough. Let me ask as a paying subscriber I was under the impression that my creations are not public. Is that the case or am I mistaken. I am just curious because the content guidelines wouldnt need to be respectful or safe for community members if its private. Like if I was making this song for myself and no one else kind of thing. Please correct me if I am wrong. I just thought the stuff I made was for me. Thanks again for listening and passing on info. Again just learning.
If it was only that simple, sadly enough there are always bad actors. Plus, Suno operates within a framework of guidelines that apply to all content, whether public or private. And besides the community safety they also have to comply with broader legal standards (Policy Review and Clarification, Customizable Moderation Settings etc).
I see, hopefully this can get sorted out at some point. This is kind of a therapy for me and would love to not have anyone or thing censoring me. But I get the complexity of laws, regulations, and all must be very expansive. Just thought I would throw my 2 sense in. Thanks.
Suno's implementation of copyright checks, although sometimes resulting in false positives, is a necessary step to comply with intellectual property laws and to respect the rights of creators globally. This approach, guided by the Digital Millennium Copyright Act (DMCA) and other relevant laws, aims to prevent unauthorized use of copyrighted material. While it can be inconvenient for users, especially when it flags original content mistakenly, Suno's team is aware of these limitations and continuously works on refining their system to balance legal compliance with user experience.
I don’t know what is going on but for the last 3 songs I get either credits refund or 9 ish seconds. I can’t finish any song anymore. (I tried but it didn’t work and cost me both a lot of credits and a lot of time.
I understand your frustration with not being able to complete your songs. There are a couple of known issues that might be affecting your song generation in Suno.
Firstly, if you are trying to continue from the end of a song and the music is dimming out, Suno might interpret it as the song ending soon and thus generate a very short clip. This is a known behavior when the music indicates a natural ending.
Secondly, there's a specific issue (bug) with the "Continue From" feature. When used, it remembers the seconds from your last input. If you are not aware of this and want to continue from the end of the clip, it might actually use the old settings (continue from xx:xx) from the previous clip. This can lead to problems like the "spin of death" or very short clips, as the old setting might be longer or shorter than the next clip.
To mitigate these issues, you might want to double-check the "Continue From" settings each time you use it to ensure they align with your current needs. If the music seems to be dimming out towards the end of a clip, you might need to adjust your approach for the continuation. These steps could help you achieve more consistent results in your song generations.
It’s neither of them . I just start a song which goes well. But every continuation fails. There are no special circumstances, it’s strange and yeah, frustrating
Can you post an example?
I’ll try
Hi guys! Every song I download its on 192kbps, is possible on 320kbps?
In the future... hopefully
This is the last one: start: https://app.suno.ai/song/617a6166-18c3-456b-9e5a-27309ec7f682
Part 2: https://app.suno.ai/song/6b47c80d-4f02-4d5e-8e82-1faa208c902d
(And one credits refunded: https://app.suno.ai/song/fff6c86d-32ea-407e-98f5-d9c7e90eb47b
okay.. let me see what it does for me.. brb
Thank you
So I keep running into a copyright popup... 😂
Trying to find a workaround that first, since it doesn't seem an existing song
Oooh hahaha, no I always check that before I make a song.
😐 it just will not let me.. ATM I cannot help you. I will let the Devs know monday
Allright, thank you!🙏
I've mentioned this several times, V3 skips intros 99% of the time and it's a major problem
I understand how this can be frustrating. It seems that V3 of Suno has a known issue with skipping intros in songs. This has been acknowledged by the Suno team. As a workaround while the V3 alpha is being fine-tuned, it's recommended to use the instrumental mode or another method to generate an intro, and then use the "continue from" feature to add lyrics. @shell yew
Some songs seem to come out as random gargled words (when using lyrics in Alpha 3)
Some of the lyric'd songs aren't completing. They cut short.
They're only a chorus and verse long so not taht long (40 seconds ish)
hey guys, today and last night, the machine's been going into this weird mode where even when I specify a female vocalist, it wants to give me either a duet with a male vocalist, or an entire song with a male vocalist. It hasn't done it in this way since shortly after v2, and I imagine it's probably a configuration setting, but really annoying
just luck or your lyrics, haven't pushed any changes yet
Anyone elses songs generating gibberish? Like, it's singing, but it's just a bunch of meaningless word-like sounds?
Yeah I have had that.
Sometimes and sometimes it doesn't even sing the lyrics
Quick follow up @neat sail about the issue I was having where half of my continuations would be new songs: This has stopped happening. It stopped yesterday and I haven't had it come back since. Dunno what you guys did if anything, but seems like it's no longer a problem, so that's good news. @summer wraith
Thank @tame latch for that one, I'm just moderating Discord.
In this particular type of AI this is usually callled a hallucination. The AI lost track of the prompt and is "off the rails". It's a bug but it's a tricky one to fix, just like it's hard to ChatGPT to stop making up fake movie names or other facts.
Can confirm, this approach works really well.
It works but we lose the possibility to generate a 2 min first part. As a help to this workaround, it would be nice if Suno would generate 2 min second parts if the first part was instrumental.
for whatever it's worth, I prepend ... before the first [verse] tags and that often gives about 2 measures of music before the lyrics start. It's far from 100% but that's half the fun of this tool anyway.
Is it just me, or are the missing variations in Part 2 (V2)? For example, I make a first part Techno and when I create Part 2, maybe 8 Times, I get different variations of how the song can proceed. Now its always more or less the same, with minor changes in Part 2. Tried it in 3 different songs now. Even if I change the complete prompts to a different gerne, the outcome is the same. Does somebody else facing with same issue?
[nvm] 🙂
Do you mean you are getting a smoother connection between songs by joining the two clips yourself, rather than clicking ... "Get Whole Song" on clip 2?
For V2 instruments this was a problem yeah, they got real boring. What people did was using lot of stuff like tags in the lyrics to try and increase diverity. [virtuoso guitar solo] [surf rock guitar hook] etc. This still works in V3, though V3 is more likely to sing the tags if you use too many. (But also V3 doesn't need as many tags to not make boring instrumentals)
omg you are my saviour - I literally have over 3000 songs generated and I have never noticed this feature lol. This is so wonderful. Wow wow wow
also facepalm 😄
No worries. That ... menu is not super obvious when it adds new options like that
It should make the songs smoother connections too
it does!!! Also, you should let the dev team know that it might be cool to have that button front and center [to the right of part x]
because if I didn't notice it over 3000 generations, then I imagine other people probably haven't also
@neat sail
Hi any solutions for the end of song that is cut in the middle of the last word ? should add outro or any other trick ?
Do you have V3? If so you can back up to the last line and re-prompt. Choose the time before the last line (or in a pause of speaking, if there is one in the middle of a line) and then prompt just the missing part.
If not, you could try prompting a blank prompt and see if it just finishes the word (sometimes it does). (You can also try this in V3, of course.)
yes i do, ok will try
@neat sail I know this is kind of unrelated, but I need to ask. I'm assuming that since you can make a part two from any song that you generate, then in theory the developers can also make a part x/continuation for any song in existence, correct? They probably just haven't released the feature yet because they are unsure about legal issues?
BUG: When ending a song with Suno, it frequently ends songs too late, leaving them incomplete at 00:59. It should begin concluding the song a bit earlier to allow for a complete ending.
BUG: Suno should treat [Tags] as what they are, namely instructions, but it should not sing the text within the tags.
I don't know the answer, so I guess I can wildly speculate? My personal guess is that it would work but probably not as cleanly or as well as a regular clip 1 -> clip 2 continue. Trying to turn raw audio/songs in neural-music-tokens-or-whatever probably isn't quite the same as just generating those right out of the model. (So maybe like a V2 -> V3 continue, pretty close?) TLDR: ???
I'm sure if it was a priority Suno could figure it out, I bet, but yeah the legal complications would be a real mess, trying to verify users ownership, seems like not worth it. Well maybe as a dev tool you would build it anyway, that's over my head tech knowledge. TLDR remains 🤷♀️
Oh okay, that makes sense. I guess it is a little bit up in the air right now. Either way, it is just insane that I am generating music for my own playlist now, on par with other music in there lol. Anyways, thanks for the help!! Appreciate it 🙂
Bug: Instrumental toggle is not persistent.
Reproduction steps:
- Go to the Create page.
- Turn "Instrumental" on.
- Go to the Library page (or any other page).
- Go to the Create page.
- All settings are retained except for Instrumental, which is now turned off.
Btw the joins are sometimes buggy, especially on longer songs, it doesn’t crossfade them properly so there’s a short (<0.5s) crescendoing “woop” in the sound
Need to submit this bug at last…
I had the same happen when I was developing the jukebox web ui, the algorithm for crossfades is harder than it seems
Is it a V2 to V3 join? (If so that may just be a limitation). If it's V3 to V3 though it shoudn't do that for sure.
No, v3 on v3
The thing is, usually it would “decode” the entire song at once. But if the song is really long it runs out of memory, so you have to decode parts separately and join them just like waves
If you download the clips seperate, is the "whoop" in the audio of the second clip?
That’s my theory, at least that’s what I had happen to jukebox back in the day
Nope, no individual “woops”:)
How old is this clip? Like more than 12 hours? Or very recent
Very recent. I’ll post now
Here. Can’t find the exact timestamp without headphones rn, but it’s in two of the quieter parts, you’ll definitely notice it
4:52-4:54
is it already reported that continuations of a v3 song still have a 1min length limit? When I create a v3 song, the first part can have a duration of up to 2mins, but when I try to make the second part, it always finishes at 1min.
Apologies if is already reported 😨
I thought that was a feature, not a bug
All continuation generations (2nd part and beyond) in both V2 and V3 have a maximum length of 1 minute. This is not a limitation but a standard setting for the platform.
Maybe setting a kind of “estimated” time limit (1min or 2) from the song creator could be a good option, even for continuations. I feel the more continuations, the more inconsistent the song is. Wrapping everything in larger parts seems to create more cohesive/consistent songs (is just a feeling I had while testing, don’t know if is it really true)
sounds like a great #💡┃suggestions-web 😄
Feel free to let the Dev know in that channel 😉
Just clicked on whole song and it produced TWO whole songs. Sadly they were identical.
I think the ultimate aim down the line is to produce whole songs in one go, but at the moment that appears to be beyond the technology as it stands.
No it doesn't, just had a part 2 of 2 minutes
That's a bug 😉
pretty decent bug
A while back, myself and a lot of other people were asking if there was a way to fix a mistake in Part 1 of a song without losing the generated style. The answer was that the "Continue from..." functionality would allow this. I've not had any luck at all with this. If I attempt to continue from the beginning (i.e. 00:01 or 00:02) of a Part 1 that has the perfect style, instruments, vocals, rhythm, etc... I get a completely different render as if I'd clicked Remix. If it was intended that this Continue from would preserve the character of the song then this is bugged
Also, I've got a song I made (5 parts) and did a Get Complete Song. After listening I decided to go back and fix something in Part 2 by continuing from Part 1 again. No matter what I do, the replacement (which SAYS Part 2 when it generates) is behaving like its a continuation of Part 5, so either the song immediately ends (ignoring my lyrics) with an ending single chord, or it does and ending where it repeats the last line of the Outro, and sometimes the first line of the new Part 2 lyrics, before ending the song.
V3 is so slanted towards female singers it continues a song sung by a man and changes the singer to a woman.
hi guys. tried to send an e-mail, but its been two days and nothing. im on a pro plan, and still no access to v3. i tried finding it everywhere, can someone look it up? thanks
Sometimes, access to V3 Alpha is restricted due to unpaid invoices. Click on "Edit Billing Details" (in subscriptions) and scroll down to review your billing history for any outstanding invoices. Side note: having the credits is not an indication your bill is paid. They give you the credits in good faith.
Any mods around? I'm getting my song flagged for moderation, but I don't understand why (no cursing in the lyrics, only mildly seductive themes)
Hey guys, moderation is hitting me for copyright when I have any short phrase followed by "this is why we drink." Not a copyrightable phrase. Not enough information to determine copyright on
This is FRUSTRATING
Just searching, there seems to be a song called "why we drink" and a podcast called "and that's why we drink.", they may be comparing generated songs to song databases and hitting generic terms like this. I was not able to use "We are the champions".
So has there been any progress in people being unable to create accounts? I see in the search for this server no one gets a response about the issue. Trying multiple accounts, multiple countries (asking people to try), and every time it's "bot traffic detected" even when going through 3 captchas and cloudflare verification lol. I have tried all three account types and used several logins for Discord, Google, and Microsoft Accounts.
Tried four different browsers, with and without VPNs as well
I understand how you feel, BlueInfinity. It's frustrating to face these issues when trying to create an account, especially after trying so many different methods. As far as I know, there hasn't been a specific mention of resolving the "bot traffic detected" issue during account creation. It sounds like you've already attempted a thorough range of troubleshooting steps. I recommend reaching out to Suno's support team via support@suno.ai for more personalized assistance.
We've tried that as well.
They are scaling up their support function and are working to improve response times, so they should be able to address this issue more effectively.
anyone elses "get full song" stuck? i tried a few times?? and i really don't want to lose this song....
If it's more than 5 minutes, it's probably not going to finish. But you can just create it again. It's also does not cost you credits for stitching them together.
My advice is then to try it again in 15 minutes. Might be a server glitch
I'm not sure if this is a bug report or a suggestion...that being said, V3 seems to 'hallucinate' more than V2 when it comes to continuing a song after 3 or 4 four-line verses...it just makes up stuff more than v2, at least in my relatively small sample of about 4 complete songs...Thanks!
Thank you for sharing your experience with V3, Imagegod. This is indeed a known aspect of V3. When the model runs out of text, it can extend the song by repeating lyrics, creating new content, or ending earlier than expected. This behavior is influenced by factors such as style, lyrics, and tags. The community has also noted that singing random words or repeating sections could be signs of the bot losing its place in the lyrics.
It's important to note that V3 Alpha is still an early version and has certain weaknesses. It can struggle with specific prompts and may not always produce perfectly mixed and mastered songs. V3 Alpha is known to be prone to hallucination, particularly when given short prompts. As V3 Alpha continues to be developed, it's expected to undergo changes and improvements.
Your feedback is valuable, and I encourage you to continue sharing your experiences to help improve Suno's services.
Thanks for the reply...I absolutely understand that V3 is in Alpha...I also understand that, really, the whole TECHNOLOGY is in alpha. I can be patient and I implore all users to do the same...that being said, I think that we're SO CLOSE to 'radio-ready' output, that patience can be difficult. I appreciate your response...Peace.
Wasting my bloody time. Two ORIGINAL lyrics get COPYRIGHT PROMPT. Bloody sick to death of this crap.
and a third...
Hi, I noticed that in V3 he doesn't finish the words completely. As if in a hurry.
And 80% have an accent in another language (mine). V2 had 15-20% accent with almost no accent.
It's missing words, it's missing lines, it's jumping from one verse to the next, it's repeating words, and lines, it's singing words that are no there.
Based on previous responses I've seen to this, it sounds like that error message is inaccurate and might not actually be caused by a copyright check. There could be something else going wrong behind the scenes with the real error message hidden from you. Apparently, they're going to try to provide better error messages in the future.
My song consists of 8 parts, but I can't generate the whole song. Whenver I click "Get Whole Song" it starts a job and never finishes it. The song contains some parts for which credits were returned (i.e. I used first 1-2 seconds from those clips because I couldn't get around otherwise). The job then disappears from the list of all songs in about ~20 minutes.
Can you use "credit refunded" clips? I've never used one before and I don't see them anywhere now in my Library but I wonder what happened if you actually tried to use part of them...before it became unavailable
Have you tried to do it in parts? So part1 and part2 make a whole song, then take that whole song then add part 3, etx? It could work..
I have used them, but only for the end of my song
...but I don't think you can merge between generated clips.. lets say you continue clip 1 to to make clip 2, then continue from 2 to make clip 3, now you go back and get whole song from clip 2... now, how do you merge 3 into that whole song? I don't think it's possible... not to mention she has 8 parts. The only choice is to redo or hope get whole song on 8th part works. Right?
id:guide I agree, it's a bot thing and not based in reality. Still very frustratin when I spend 30 minutes composing some original lyrics only to be told by a machine that they are not mine.
Last 6 sections have all generated silent/
Those should be refunded, imo
It's an interesting test case, for sure. So that is why it's good to test my suggestion.
Just send an DM to @muted flame so he can look into that. Send him the urls of all the clips
I can see it might work.. get wholesong in smaller steps... then combine them offline on DAW
hmm.. That is strange. Part 1, Part 2, Part 3, and Part 4 and no longer in my account. They are gone somewhere even so I didn't delete them
Check in Library
did you check your library?
Says credits refunded now, SO the system must have spotted it 🙂
now some wiered screaching sound that the start LOL , I can never get a full song out of this 🙂
been playing for weeks
I do it all the time. It has more memory for the melody that way 😉
Yeah me too.
Yes, they are not there. I suspect, the reason why it could be is because I reported the other song which was generated and it removed 2 songs instead of 1. I.e. I generated Part 1, Part 2, Part 3, Part 4 and each time it generated 1 normal 50-60 seconds song and another one which is 10-12 seconds (without completing even the first verse). So, I reported the other song.
Now, I suspect, that if you report a song it will remove not only this song, but also the other song as well. Not sure if that's really the case, but none of Part 1 - Part 4 songs are in my library.
You do know the option that you can continue from anywhere right? That might help?
when you said "report it"... did you mean thumb down? all my thumb down clips are still in Library
yer but that would roll back on the tune its kind of changed as it went along and it works
I clicked 3 dots -> Report -> Bug. And described that the generation of the song is stopped at the beginning of the verse and credits were not returned for that song.
it shouldn't but there is a know bug atm. Always double check the timestamp. ATM it doesn't reset with the next continuation.
I never had that made my generations dissapear as far as I know
oh well another wasted hour 🙂 all are coming back blank and credit refunded .
Interesting. I did not even realize there's that option 🙂
I'm not sure why those Parts are gone other than the reason that I reported the other part (another version). There could be some other bug which I'm not aware of. That's sad 😦 I liked those parts...
You can check your trash.. go to library, then besides the "public" button you see 3 dots. Clip on that to go to your trash
No, it's not in the trash. It shows "Your trash is empty." 😦
Aww.. really sorry about that.. I don't think it can be fixed then. 😦
But I would like to give you some advice.. Next time when you get those super short clips, make a whole song of what you got. That way you cannot lose it
You can simple just continue from that whole song
And it has more memory as well for your melody
So it is not a solution, more a painful lesson... and I hope this doesn't discourage you
Thank you so much for your help! Yeah, that's a painful lesson. I had to download each generated part and not relay on the system to store these parts... I will try to generate the song again later and hope it will be similar to what it was. Thank also for the tip to generate the whole song more often. Maybe it will help.
Interesting fact. I clicked to generate "Get Whole Song" from "Part 5" and.... it's gone 😅 It seems "Get Whole Song" now works as "Remove the song" in my account. Screwed up again 🙃
I think it would be wise if you share some of the clips with @muted flame so he can take a deeper look into that.. this is getting more than just weird lol
Anyways, I sent an email regarding this issue. Maybe they can check my account to see what's going on..
don't forget to add the urls of the clips.
click on the white arrow from the song to get that
In some songs that I try to create, I add the lyrics and when the song is created it only plays the solo and the voice does not come out.
https://app.suno.ai/song/845dd53b-696a-4b2f-a92b-801bd7b8cfd3
https://app.suno.ai/song/9779a62b-23b5-45be-bf34-3155748eb8a1
https://app.suno.ai/song/a53aed4b-4dda-4efe-821e-c561e86ff1a9
https://app.suno.ai/song/24980488-269f-4394-893d-a3a91d9e8f83
https://app.suno.ai/song/220d825d-b370-4dd3-a263-58eb36fc97c1
https://app.suno.ai/song/8bf5de2c-7b87-40c0-944b-d63d355af48a
https://app.suno.ai/song/f47d5299-51f1-41e3-bb85-725b656bf031
https://app.suno.ai/song/ed680668-498a-41a9-a41a-f62d0a654581
All of these are "Part 6" trials. This is the earliest part I have. Parts 1 - 5 are gone from the account. I can share Parts 7 - 9 if needed as well.
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
please check our community wiki about this: https://www.suno.wiki/faq/making-music/why-cant-i-hear-my-lyrics/
@muted flame let me know if I can share anything else to figure out the problem. These are all Part 6 songs, none of them completed the "Outro" and for some reason Parts 1 to 5 are gone from my account (thus I cannot get the whole song). For Part 7 I just used this part https://app.suno.ai/song/8bf5de2c-7b87-40c0-944b-d63d355af48a but with custom time (from 0:10).
Could be that it's related to V3 model, but it's strange that the songs disappear from the account 😦
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
Okay, it's strange. Part 5 was in my account. I clicked "Get Whole Song" and it was removed from my account. After awhile I see Part 5 is now back in my account: https://app.suno.ai/song/856bdf97-5fc5-475f-a946-848c03c862a3
However, Part 1-4 are not there.
Electro-Pop, Dark Pop, Gothic Pop, Indie Electronic, Synth-Pop, Female voice, Sexy voice song. Listen and make your own with Suno.
We should start a lost and found stand lol
I have many versions of part 2 and part 3. Is there a way to get the whole song but choose which ones you want?
I keep getting a message telling me it can't generate for copyright reasons when I try and generate the last part of the song I'm working on which is just a repeat of the chorus I've been using the whole song and a simple outro. 🙄
Hey guys, the following is not a line that's specific enough to be copyrighted anywhere. But I'm getting dinged by the moderation system again
Oh-oh, oh-oh, oh-oh (You don't know you're doing it)
Oh-oh, oh-oh, oh-oh (this is getting old)
I'm no expert (at all), but you might try to fool around with the lyrics...or you can try signing out/in. For example, try replacing oh-oh with oo-oo...it may not be optimal, but it MAY be a useful work-around (sometimes). It's a wonky bot...but it can produce some real magic...Peace.
Yes, but then the machine hallucinates less
which is not what I wanted here
it's all good, I got past it
I've noticed that I get the "system error" when the Style of Music box has no character space left at the end. Removing just a single character allows me to create.
guys, the moderation system is on the fritz, and it's making songs impossible to finish
please look at it
it's blocking all small pieces of text
I don't understand why it needs to work this way, there's never ever going to be copyright infringement in three lines
Currently, Suno is researching/developing a feature known as the "paint in" feature, which is not yet available but is highly anticipated. This feature would allow users to select specific parts of their music, like a bridge, chorus, or verse, and regenerate just those portions without altering the rest of the track. This would provide more precise control over the editing process, giving you the flexibility to choose which versions of different song parts you want in the final composition.
In the meantime, to combine multiple parts into a whole song, you can use the "Get Whole Song" feature. This feature stitches together all the clips in a chain into a single song. Note that while using the "Get Whole Song" button does not use credits, each continuation of a clip does use credits, just like the first song in the chain.
I understand your frustration with the moderation system incorrectly flagging your lyrics. This is a known challenge and the Suno team is actively working on improvements to address these issues. In cases where lyrics are mistakenly flagged, a workaround could be to make slight changes to part of your lyrics. This has been a necessary step for others as well, and while it's not ideal, it can help bypass the current moderation limitations.
Remember, Suno is continually evolving and strives to refine its systems for a better user experience. Your patience and feedback during this process are greatly appreciated.
Hi, would you please be so kind and explain that option? The only thing I know is, to continue a song when its done from the endppoint. Thanks!
this should explain it 🙂
Currently, Suno does not have a feature for generating parallel musical notation alongside audio. Suno focuses on audio generation, where the melodies and music created follow certain patterns influenced by style, lyrics, and tags used in the prompts. The team is continually working to tweak these influences for the best outcomes.
Developing a feature that generates musical notation could indeed provide users with more control over the music generation process, similar to how photo editing software allows for fine-tuning. This would be a significant enhancement, especially for users who wish to have a deeper involvement in the music creation process, beyond just generating audio. However, as of now, there's no indication that such a feature is in development or available in Suno.
But in the future, anything is possible 😉
I am FED UP of creating original lyrics with prompts only to be told they are NOT original! Here are the two from today:
1-2-3-4
[intro guitar riff]
[verse]
You’re never in
You’re always out
When I call
There’s no doubt
That you’re playing dirty games
That you’re playing dirty games
[chorus]
You know how to lose it
You know how to use it
You’ve got to choose it
Make up your minds or make up your face
[verse]
You’re never in
You’re always out
When I call
There’s no doubt
That you’re playing dirty games
That you’re playing dirty games
[chorus]
You know how to lose it (prostitution)
You know how to use it (prostitution)
You’ve got to choose it (prostitution)
Make up your minds or make up your face
[chorus]
You know how to lose it (prostitution)
You know how to use it (prostitution)
You’ve got to choose it (prostitution)
Make up your minds or make up your face
1-2-3-4
[intro guitar riff]
[verse1]
I wanna go home I got no place to go
Turn on and listen to my are ay dee eye oh
I listen to the long wave I listen to the rest
Tune to vee aitch eff for my dads’ request
[chorus]
I’m looking for something better
I’m looking for something better
I’m looking for something better
[verse2]
I hear the top twenty ringing out
Rock’n’roll power pop twist and shout
The number one gets more than it deserves
The whole damn thing’s getting on my nerves
[chorus]
I’m looking for something better
I’m looking for something better
I’m looking for something better
[bridge]
are ay dee eye oh
better!
are ay dee eye oh
better!
[building]
are ay dee eye oh
better!
are ay dee eye oh
better!
[verse3]
There will come a time when I’ll be rich
I’ll play all my favourites without a hitch
All the little DJ’s will do what I say
I’ll rule the street I’ll rule the town I’ll rule the world
[chorus]
I’m looking for something better
I’m looking for something better
all this work time wasted.
Hello, i have a bug in my account I have the better subscription and I can’t make more 2 songs in the same time and I don’t have v3 version ?!
So I'm thinking it's not the moderation system. I think it's a notation error. All of the other technical errors seem to have gone away in favor of the copyright error, and I managed to fix the problem by adding a line to the verse, which should not have worked if was the copyright moderation.
song came out great though https://soundcloud.com/holographicgf/over-you
I absolutely love writing bitchy country songs
Lyrics:
The last few weeks have been amazing
After saying our goodbyes
Sometimes things aren't meant to work out
we move on with our lives
I don't thi
I understand how you feel, and I'm sorry to hear about the difficulties you're experiencing with your Suno account. It seems like the issue you're encountering, particularly with not having access to V3 Alpha despite having a better subscription, might be related to billing issues. Suno has noted instances where access problems were linked to billing discrepancies. I recommend checking your account page to ensure all invoices are paid. You can do this via "Subscriptions", then "Edit Billing Details". If this doesn't resolve the issue, please reach out to Suno's support team for further assistance.
My id is : iammikeia
I am a guide. Just check "Subscriptions", then "Edit Billing Details".
All the cases I have seen, have a billing issue. And only YOU can fix that 😉
Please check "edit billing details" as I said @thick raptor
please ask your question in the right channel. This is meant for bugs. Not #💬┃general-chat or #📚┃suno-school questions
Step 1. Made an instrumental clip
2. Continue from song on instrumental clip
3. Change advanced to simple mode, enter prompt and continue
4. Only generates instrumentals even after continuation track is cleared from advanced, cannot add lyrics in advanced either.
yeah one of my companies has a "beats store" and it would be neat to integrate it directly in the app lol
I am actually having a lot of problems with it knowing how long the song is in safari until a long time after it has finished generating and has been refreshed multiple times
i can't fast forward or rewind in a song for maybe 5-10 mins after generating
I can't use /chirp
I get 14 seconds (when my lyrics should be going close to one minute), credits are not refunded, the second one got 6 seconds credits refunded,
the 14 second clip was good, but I was expecting the song to be finished and now I need to do another round of 10 credits.
it's not on my account connected to discord, I prefer to stay anonymous, if I can send a DM to get it taken care of. Thanks.
I recommend sending a direct message (DM) @muted flame for personalized assistance. They should be able to look into your account and take care of the issue regarding the short clip and credit refund.
Since this week, the V3 has not been running smoothly. Aborts in the middle of the song and with continue from complete verses are ignored or single words or verses and chorus are mixed. It's less fun on Friday when the V3 ran much better
thanks
in english: it mispronounces the word "geek" as "jeek"
in italian: it mispronounces the letters "che" when it's found in the middle of a word, which are supposed to sound like "keh". The sound it's currently making is how you write "ce" in italian.
For example "cachetta" (meaning a little poop) it's pronouncing it as if it was written "cacetta", from an italian linguistic prospective.
there is so much autotune-effect on vocals...
The autotune effect you're noticing in the vocals generated by Suno reflects the current style trends in music, where autotune is widely used. Suno's models learn from existing music, so they incorporate these trends into their creations. If you're looking for a more natural sound, trying styles with minimal effects like unplugged music or simpler genres like hip-hop might yield better results.
Yes, that is all true, but I was making German lederhosen ompabompa 😄 And it creeps into every style.
I believe the issue stems from sound saturations. Whenever there are high pitches, the sound saturates. The overall compilation is not clean, with crackling sounds. Is it possible to achieve better sound quality results?
In the future anything is possible 😉
it's mostly notable whenever the melody shifts notes too quickly
So, somehow I mostly get only 17 sec, sometimes also 11 sec clips created even though my "verses" need at least 29 sec. I'm currently at Part 2. I would understand it at Part 6 or 7 but right at the beginning, it's crap. Does anyone have a solution for this? It's the same case with every song.
I understand your concern, pescholino. It's indeed frustrating when the generated clips are shorter than expected. This can happen for a few reasons. Firstly, ensure you're providing enough lyrics or content for the AI to work with. If the input is too short, the AI might not have enough material to extend the song. Also, try adjusting the continuation point if you're continuing from a previous clip. Sometimes, setting it a few seconds before the end of the previous clip helps. Importantly, adding more details in the "style of music" field often leads to longer generations, so give that a try. It's a bit of trial and error, but these adjustments can often lead to better results. Remember, AI music generation can be unpredictable, so keep experimenting with different styles and inputs.
Ok, thank you. I will try out a few new things.
After many experiments (and credits spent) with V3 something that I notice that fails and is quite frustrating for my workflow is the consistency in the BPM speed. It never maintains a fixed speed (example 130bpm), there are continuous micro changes (sometimes more obvious) not predictable that makes it impossible to work with the generated track in a DAW afterward.
Which DAW are you using, if I may ask?
Studio One 6
I use stem extract in flstudio, but you'll get better help in one of the channels that 3Daizy mentioned.
I can see why this BPM inconsistency issue in Suno V3 is causing frustration, especially when trying to work with the tracks in a DAW like Studio One. A practical solution I've found is to use a tool like Serato Sampler to determine the exact BPM, which is often a non-whole number (e.g., 130.23 BPM). By identifying the precise BPM, you can adjust your DAW settings accordingly, making the tracks more manageable and compatible for further production work. This approach, while not perfect, can significantly improve your workflow with Suno V3 tracks.
Thank you I will try
Some songs generate with no lyrics even though I did know select instrumental.
Did not*
@low cypress
This contains some knowledge from our community about that subject
Thanks for the fast reply I’ll try. Would it be a bug if one song generates lyrics and one does not from the same generate prompt?
Scratch that. Just read.
Appreciate the help.
This does not work. Serato Sample is also unable to detect the exact BPM of the tracks generated by Suno. It is impossible to sync with other beat tracks in the DAW to use it in production. ...a pity really. I hope this will be fixed in the near future.
Weird.. it works for me.
Serato Sample (like most tools for this) makes a reading of the initial bpm of the track. The problem is that the tracks that Suno generates have micro-variations of BPM continuously throughout the track. So it makes it impossible to work with the Suno track.
When I find some time I will make a video how I do it.
Gracias! estaría genial!
This is how I do it
any mods here any idea why I emailed on saturday to customer support on sat and still no reply? Is there some faster channels?
I got replied the other day to an email I sent on Dec 26. I guess they’re überswamped
Man, a computer is now writing music that makes people cry. How fast have we forgotten how amazing and unbelievable this is and switched to being frustrated by microvariations in bpms?
V3 seems a bit better but with female vocals when it sings high it is ear piercing HIGH and can't do much with it after.
#kpop
v3 and all the new feats disappeared and I am still a premium user.
still there for me
hello. i wrote to the support
since 2 weeks, nobody is answering
because of the bug on the Suno.ai page
@stone isle managing high-pitched female vocals in Suno V3 can indeed be challenging, as the alpha version is still refining its vocal generation capabilities. To mitigate the issue of overly high pitches, consider specifying the vocal qualities in your prompts more precisely. Use terms like "alto," "mezzo-soprano," or "soft female voice" for a more moderate pitch range. You could also describe the desired vocal quality using terms like "gentle," "smooth," or "warm" to guide the AI towards producing more balanced vocals. As V3 is in the development phase, experimenting with different descriptors and providing feedback on your experiences can be instrumental in helping the Suno team enhance the model for future updates. Additionally, adjusting the music style to one that typically features more controlled or lower-pitched vocals can also influence the overall vocal output of your creation.
I asked for Gregorian chant and got two women https://app.suno.ai/song/a948b676-9030-4505-bf00-153772229e74
Gregorian Chant song. Listen and make your own with Suno.
This situation highlights some of the challenges and limitations inherent in Suno, particularly in their current development stages.
To improve the accuracy of the results:
-
Be Specific in Your Prompts: When requesting a particular style like Gregorian chant, provide as much detail as possible in your prompts. This might include specifying the type of voices (e.g., male choir) and other characteristics of the style.
-
Use Relevant Metatags: Use them to guide the generation towards the desired outcome. For example, tags like [male choir] or [monks singing] could be more directive for a Gregorian chant style.
-
Experiment with Different Inputs: AI music generation can sometimes be unpredictable. Trying different variations of your request or adjusting the wording can sometimes yield more accurate results.
Keep in mind that AI music generation is a complex field, and the technology is still advancing. There might be limitations in accurately replicating specific music styles, especially those that are less common or have very distinct characteristics like Gregorian chants.
hi,Cancel my purchase, I don't know how to cancel, help me
Please contact billing@suno.ai
i contacted them, they don't response
The team is very very small, and I guess they all do a little bit of everything
https://app.suno.ai/song/c296893b-8a1a-40c4-8468-cec0f8fec227
Can't hear the Anthem Emo riffs I requested
I hope at least the Beta is as good as Alpha was on release, it could blend the most nonsensical things. I'm counting on you guys, especially the dev team. I know you've seen me here before.
Dissonant and out of key notes on this song:
https://app.suno.ai/song/1a77cdc5-92bd-420f-96c7-eab651d9e9c2
I keep getting songs without vocals.
Alternative rock, pop punk, clean female voice, catchy, intense, restless song. Listen and make your own with Suno.
Thank you so much.
unable to remix
for whatever this is worth:
lolmarketingexecs
I'm not sure augmented chords or even things that are out of key are really a bug so much as not what you were hoping for. Getting unexpected results is one of the points of these AI tools.
and ambient music has a lot of dissonance and altered chords a lot of the time. Very much of it is even atonal. I mean, we're talking about Brian Eno's genre
the support is not responding to my e-mail, what can i do?
wait because they're a small team of human beings who are very busy
since 3 weeks...
i get no points, sendet them the contract and payment letters, i get no points, since i payd
i had only 250 tracks available
and the points i should get to make it 500 are not stacking
since 4 weeks around 0
and now again
Thank you
hey there, can you DM me your account email and someone will look?
@muted flame i wrote you in dm
most of my songs are slowing down now after i continue them from the first part, they will slow down toward the middle of the lyrics prompted like a record
Hi, the idea of the new library is very good but unfortunately it is not accessible with my screen reader. I work with Mac and voiceover.😔
What would it take for them to make it accessible for you?
In the previous version I could do everything without a problem. Now I can only read the title of the song but I can't access the other options like share and more anymore. I don't know how you can change the design so my screen reader can read and access it.
This makes me very sad but I know it's not on purpose. Would they be happy to have someone to check the accessibility for new functions? I also know blind Windows users who would be happy to do this so it can be checked for both windows and Mac.
I am sorry to hear that! I will take note and let the Devs know.
Thank you very much. Can I DM you an email address where I can be contacted in case they want someone to look for the accessibility?😁
Feel free. I cannot promise anything (I am a volunteer guide), but I will get it across, that is the least I can do.
We will look into this! Are there any other options that are inaccessible?
Thanks @low cypress ! Could you actually DM me your email? (Although 3DAizy would have dmed me anyway 🙂 )
Thank you so much! This is the very first time something hasn't been accessible. I can't access the library and I also read you can make playlists now but I can't access that either. Thank you so much again
Is the download function working for you? I heard from other people who use screen readers are not able to.
I can't access the download button anymore. I can only play and stop the songs and even there it slips around all the time. You might get different feedback from people who use windows and Mac
Thank you for sharing that.
My credits weren't refunded with an 8 second generation.
that's probably like 8.1s 😦
in any case, we will tweak this to be a little more reasonable soon
Oh ok. I think 8 seconds is a bad number because it's expected to do things in 10s normally.
Hello all. I usually get about 10-20 "good enough" songs out of 2500 credits. The songs need editing of course but its usually ok. Since v3, I have not been able to get 1 song out of 2500 credits. Is it possible that v2 quality is getting taken away somehow for making v3 better? I tried both versions and neither worked for me. Is there an estimate on how long this will be?
https://app.suno.ai/song/2610e006-0716-4f8e-8bc7-2cbfecbe84bf
https://app.suno.ai/song/5781d69f-a936-41bf-9111-2592d4a6db60
It keeps skipping to "invisible" line. I even reordered and had the 4th line sung first.
v3 is a test version, why would you not expect it to have more bugs and issues
they wouldn't take away from their commercial version to power the test
that would be crazy
probably not even running on the same resource stack
Yesterday I was complaining that some of my songs disappeared. Well, apparently it's me being stupid and not seeing that there are multiple pages in your library. After clicking on that I found all my songs. Sorry for false alarm 😅
Playlist: One can add the same song multiple times to the same playlist. Bug or feature?
oh sorry, wrong channel
Haha. Are you serious? Im talking about v2 dude.
did you read what I said after it when I realized what you meant?
If I told you that Chirp V3 (alpha) generated a song that have a duration of 3:45, do you believe me?
(but with something veeeery strange but kinda interesting about how it tries to stick the pieces back, from it's model I think?)
hardcore-frenchcore-uptempo-gabber song. Listen and make your own with Suno.
I got 7:13 once
Training a model is very computationally expensive. I can't imagine that v2 ever changes regardless of any new checkpoints that they add. There are other parameters like sampling steps or whatever that could get re-tuned but I doubt that this would happen once they get it dialed in. You may just have had a string of bad luck. It happens to me too.
Highlighting previous parts of the song from continuations and full songs seems to stopped working.
Using "Marshall Mathers" as a prompt got me a song called "The Eminem Express" with the lyrics to Lose Yourself
Using "Slim Shady" as a prompt got me a song called "Shady's Back" with the lyrics to Without Me
Put in "Killer Queen" as a prompt, got the lyrics to "Killer Queen"
and yet they ban my original lyrics lol
cannot locate the button for stitching it together and the whole thing is strange
V3 seems to do a lot of less than a minute now.
"Song Description contained artist name" 😐 -> if an artist name is common word, we would never be able to include it into the lyrics?
I can see it up the top above Rename "continue from this song". Also why V2 when you have the V3 option available??😳
The new update is a little weird on iPad, I have stuff in the wrong playlists and my playlists jump around when I try to scroll through them. The menu to continue or remix doesn’t always come up because another menu for adding to playlist comes up instead. Just have to keep clicking until the right menu comes up, seemingly for no reason.
Alpha 3 no longer continuing song. In the past it has carried the lyrics over, now it doesn't and I had to paste the lyrics in myself. Then I put my usual prompt at the top which is "Continue the below song by adding a new verse but keep
the chorus the same. Maintain the vibe of the song." > Press the lyrics button and it just retains the original lyrics. The continued song does not get produced and my prompt is subject to moderation.
V3 alpha produces far better results than V2. I have had 99% success rate on my 2500 credits. However today the continue song option stopped working.
for people saying they can't get good results on V3 alpha, I don't know what they are smoking. https://app.suno.ai/song/6d655572-0ef8-43fd-a51f-a0e46bc708d6
fast-paced rap aggressive song. Listen and make your own with Suno.
constantly being flagged for moderation? This is great software for horror music songs for my series. why is everything being flagged now?
(Verse 1:
Howling winds of ruinous fate
Swept through the drowned city's gates
In hurricanes wake, madness reigned
Vincent's descent, sanity waned
Gail's unholy birth, a brain encased
In rusted steel, hunger unchaste
Feline flesh, her cravings grew
Jinx, the catalyst, chaos ensued
Chorus:
Feasting on whiskers, disemboweled fur
Gail's rampage, an ungodly purr
Chefs kidnapped, sauces plundered
As the city's culinary soul sundered
Verse 2:
Hotboxed in his lab, Vincent's high
Spawned rubber chickens, an affront to the sky
While Gail's culinary crime spree raged on
Kregon and Jinx, their mischief, a hellish dawn
Twerking before captive cooks
As pots of horror bubbled and shook
A knock, the cat's devilish grin
The absurdity escalated, no end in sight
Chorus:
Feasting on whiskers, disemboweled fur
Gail's rampage, an ungodly purr
Chefs kidnapped, sauces plundered
As the city's culinary soul sundered)
why is this getting flagged?
im on a dead line.
I'm about to bust out the piano and guitar and write this myself.
any moderation peeps on here?
I'm currently using the website i subscribe.
My continuation of song prompt is being flagged for moderation.
This is why. Same song, just remixed and it cuts off the last word to "miss":
https://app.suno.ai/song/189f487c-f48a-4a31-90bd-022a23389e58
95% of things I generate a excellent, but cut off the last word, or the last line.
Edited for wrong song.
fast-paced rap aggressive song. Listen and make your own with Suno.
I've rewritten this lyrics 5 times still get rid of the moderation it's music, it makes no sense.
Why is womb and katrina flagged? fix that!!! I'm in new orleans that's absurd. Comon music it's music it's art it's creative outlets.
"woom"
"kuhTREEnuh"
It'd be more understandable if the lyrics were generated by Suno that would have this kind of filtering or restrictions.. but if users provided lyrics... Suno shouldn't care
probably to do with some woke nonsense.
I got it to work rewriting it womb and katrina had to be removed
update on continue bug. I've tried every song in my library about 13 songs, lyrics and style don't carry over. Update I just copied and pasted lyrics manually and the style manually and my prompt "Continue the below song by adding a new verse but keep the chorus the same. Maintain the vibe of the song". Hit generate lyrics and this time it worked and produced a continued song. Yay. First success in 10 attempts. some progress. Maybe just a glitch for an hour or so?
Maybe a glitch, there were some updates today. It's not that uncommon for a single clip to have trouble continuing, sometimes it's a problem in the clip. But it 100% should not be a thing that happens with any clip you try. Also check the 'continue time' feature if you are using V3, maybe it was stuck on a strange time.
I'm pretty certain I got part of a real song here at 24 seconds onwards, even though I didn't try to https://app.suno.ai/song/47489bb4-344f-4743-99dc-7ab175786cff
Yes once again now seems to be working correctly again. Thankyou Suno.
Continuation of song bug. Most of the time it only plays the verse even though the chorus lyrics are on the screen. My last attempt it finally sang the chorus but missed the last word. V3 Alpha.
The cutoff timestamp now resets to full song at every closing of the custom tab.
cant add songs to playlist if I use the link, any help»
but the message that shows is 0 clips added to playlist, and then its not there
are you trying to add someone else's song to your playlist?
that might be a bug if they allowed and if they don't allow it, then the menu option shouldn't be there... either way, sounds like a bug to me
@low cypress Is yours also unusable now?
It's working for me. Might be a fluke
@neat sail I think the model for generating Valentine's Day songs is great, but it hasn't been working lately.
My Suno is behaving in totally strange way since yesterday. Whenever I click move to trash it remove song from trash and whenever I click remove from trash it move song to trash and also when in trash when I click delete it permanently it just stuck there forever and song is never deleted! Is there any solution to this strange and exasperating problem?
I have to click away this message every time I want to change my text. If I have to acknowledge the message every time anew, wouldn't it be better to have a permanent message on the website, visible to everyone, without the input field being blocked or always having to accept a message first before i can change my text?
"Please accept our terms to make a song with custom lyrics
Suno is designed for creating original music. Please confirm you will only submit AI-generated lyrics, original lyrics or lyrics to which you otherwise hold right to continue."
Sorry, the Valentine's day site is no more. But the same model is available as V3 Alpha for non free plans. (And later on free plans too.)
You should only have to do it once, so that's 100% a bug!
Any solution for my problem?
Try control-R refreshing your browser page, or forcing a reload in some way. Not sure otherwise
Didn't worked a bit, I even tried a different browser. It appears that the problem is at AIs end and not at my end!
And the worst of all is that there isn't any option connect help team directly!
Any genre containing the word freestyle does not work, eg: 'freestyle rap', 'freestyle latin' 'freestyle pop', etc. Freestyle is an entire genre itself. So, just because someone's artist name is 'freestyle', suno should not mark everything that contains the tag/word 'freestyle' as an artist. https://en.wikipedia.org/wiki/Freestyle_music
Freestyle, or Latin freestyle (initially called Latin hip hop) is a form of electronic dance music that emerged in the New York metropolitan area, Philadelphia, and Miami, primarily among Hispanic Americans and Italian Americans in the 1980s. It experienced its greatest popularity from the late 1980s until the early 1990s. A common theme of fre...
I do not have a dropdown for V3 😦
V3 Alpha is for non-free accounts only. If you do have a pro plan, then check the members area in the account screen to see if maybe a past payment didn't go through or there is some payment error.
I set up a new payment and still nothing I had premier plan and switched to a pro.
Did you have an error in there? if the payment had failed and you just fixed it now, it may take a little time to get V3 access. (not sure how long, but minutes I think?)
When did your premier plan officially end? Because as far as I know your new Pro plan will start on the renewal date of your original Premier subscription.
Please check via subscription and click on "edit billing details"
Fixed. Thanks all.
I know we had a bug that it kept the timestamp you filled, even if you switched to a new song to continue from. So I guess they tried to fix that. Can you explain your problem exactly with it now?
im not allowed to click create anymore?
i put all the information in to create a song but it won't allow me
y?
Have you filled in all fields? (lyrics, style of music, title)
Also check if you still have credits
Yes, I experience that often too. The problem occurs for me whenever I manually set the timestamp, for example, to 00:20. If I then decide to "reset" the timestamp by deleting it, the "Create" option becomes unavailable for me.
Is SunoAI in generally a GPT/LLM?
Yes, SunoAI functions as a large LLM (Language Model).
It doesn't for me. Can you try again and give me step for step how you are doing this?
Minor cosmetic bug that I discovered by accidentally having something holding my ENTER key down - if you click on the "next page" arrow and hold ENTER down, it will continue advancing the page number beyond the last page (I ended up on page 1000-something before I realised lol)
Anyone having an issue with this
Any time I try to update my billing cancel my plan it is stuck like this and I can't upgrade or anything
I understand your frustration with being unable to update your billing or cancel your plan. To address this, here are a few steps you can take:
- Verify Card Details: Make sure that the name, card number, CVC, and expiration date on your card are entered correctly.
- Check with Your Bank: Sometimes transactions are blocked by the bank for security reasons. Contact your bank to ensure they aren't blocking the transaction.
- Confirm Available Funds: Ensure you have enough funds in your account to cover the subscription fee.
- Email Suno for Assistance: If the problem persists after these steps, it's time to contact Suno directly. Email their billing support at billing@suno.ai with your most recent invoice number and a detailed description of your issue. This will help them to investigate your account and provide further assistance.
Taking these steps should help in resolving your billing issue and allow you to manage your subscription effectively.
Yeah I sent an email about it just waiting for a response now
I updated my payment details already
I've sent this also for you to billing support. Hopefully they can pick it up soon for you.
Thank you
It’s in the video. So basically once you set any timestamp, then switch to simple mode (to have more space) and switch back, the timestamp resets to full song.
I had similar issues when I tried Style: Paso Doble, it says some artists name "Doble" - they said they'll fix it but I haven't tried again
I think they changed that.. since it does work now (paso doble), but it doesn't seem to recognize the genre in general much. 😬
oh hmm
maybe it wasn't trained to recognize that genre/style.. its' a shame bc I wanted a song
Please send an DM to @muted flame. He has let me know he that he will personally help you with this.
Greetings, although I have enlarged the screen, the download button and other buttons remain below
By the way, I have a professional membership, can I get a role in December Dec
I see why that's frustrating. To achieve longer continuations, consider these tips:
- Check the content you provide: Ensure your input lyrics are sufficient for the AI.
- Adjust the continuation point: Use the 'continue from' option to avoid areas of silence or fade-out in the original clip.
- Detail your style input: More detailed information in the "style of music" field often results in longer songs.
- Experiment: AI music generation can be unpredictable, so don't hesitate to try different styles or inputs.
If issues persist, remember that the Suno team is continually working on improving the AI, and your feedback is valuable in this process.
Testing these tips
Hang on
Well, I can't complain about the quality any longer. It seems pretty in line with what it was when it released, which is nice, but yeah, the continuations are pretty short for this specific song, for some reason.
It's intriguing that this issue is specific to your song. Short continuations can occur if the AI does not have enough material to work with or if it's influenced by the ending of the previous clip. Try adjusting the 'continue from' point to a different section of the original clip where the music is more consistent. This might help in getting longer continuations.
Sure will
I've told Suno to continue from 0:33
After that, there are vocal hallucinations
what's your music style
Pop Punk/Anthem emo
try adding in more.. that's a bit short
Oh, hang on
natural vocals Virtuoso Anthem Emo, Alternative Emo, Orgasmic Athmosperic Pop-Punk-Post-Hardcore
This is what I put in the style box
Quite the mouthful
yeah.. that's pretty decent
are you willing to share the url? I wouldn't mind to see if it continues better with me ;P
Sure, just a minute
Ultra fast Virtuoso Anthem Emo, Alternative Emo, Orgasmic Athmosperic Pop-Punk-Post-Hardcore song. Listen and make your own with Suno.
needs to be instrumental. right?
No, only this beginning
This intro
After that, there are lyrics
I'll try continuing from about 27
okay I will put in some dummy lyrics then
Ok
When start around 24 seconds it giving me decent clips it seems.. still rendering but will share in a few minutes
Okey dokey
https://app.suno.ai/song/fc7ba09b-033f-48e6-bdd4-b23c707395fa
THis is what it came up with starting at 27. I really like it, but too short, unfortunately,
natural vocals Virtuoso Anthem Emo, Alternative Emo, Orgasmic Athmosperic Pop-Punk-Post-Hardcore song. Listen and make your own with Suno.
You might have to put in more lyrics then
natural vocals Virtuoso Anthem Emo, Alternative Emo, Orgasmic Athmosperic Pop-Punk-Post-Hardcore song. Listen and make your own with Suno.
I now have around 30 seconds but more because I don't have more text
natural vocals Virtuoso Anthem Emo, Alternative Emo, Orgasmic Athmosperic Pop-Punk-Post-Hardcore song. Listen and make your own with Suno.
Not bad at all
I might just try that
Well, if it doens't work, at least I'm happy Suno is pumping out some really nice stuff
They sure are!
you cant use suno on copilot ever again after you generate a few songs with it
Anyone else experiencing song after song missing your lyrics (V3 on the site)? Like 5/6 of my last generations were 100% instrumental (option not selected). Last night it was fine
maybe this helps: https://www.suno.wiki/faq/making-music/why-cant-i-hear-my-lyrics/
"Get a full song "is taking forever. Any ideas?
I understand your concern about the "Get a full song" feature taking longer than expected. It appears that in Suno, if a song generation process is stuck for over 5 minutes, it's classified as being permanently stuck. In such cases, the generation process will not complete, and the song will eventually disappear from your queue. It might be necessary to start the process again or try generating a shorter clip if the issue persists.
It worked perfectly 12 hours ago, and usually, the entire song I try to generate takes less than 4 minutes. will try again. Thanks
It's not uncommon to experience fluctuations in performance with Suno V3 Alpha, given its developmental stage. These ups and downs are part of the growing pains of pioneering AI technology. If you find these issues too frequent, remember that you can always use the more stable V2 version for a more consistent experience.
hi, during continuation usually, but not necessarily, also the first generation too, half of the time, it's skipping some words imn the lyrics. has anyone had the same problem like this? sometimes it will skip a whole line, and sometimes will mash up the bbeginning of one line and the ending of the next line. I'm very plesed with the results quality of V3 but such bugs, I'm wasting credits like never before. because evén a small skip of the words, ruins that generation, and i have to re-do, and when it's some certain moment, i don't know depending on what, it will compleely stopo generating mistake-free results, and every result will have some sort of mistake or oddity, like words skipped, spelled wrong, for one single line the voice will change to another gender (for example, when a female voice is generating the song, suddenly. for one or two lines, it goes to mle voice then back), often it will give one of the results withot singing, just instrumenal by its own decision, there has been many times very veryy short results generated. and maybe 7 times ot of 10 attempts, the option to continue creating the nexxt the clip from the selected time of the previoous result, doesn't really seem to care the preferred time to contine from. it will just continue wherever it wants. I got mch better results in terms of the qality of the sound, but the problem is I have spend at least 15x more credits trying again and again until i get a reslt wih no mistake or a missing part.
So this is a thing that happened, two minutes of total silence https://app.suno.ai/song/473a2ecc-0904-451b-891c-1d4cec2837d3
(Btw, before someone mentions it, let me st say.. 😬 I wanted to share this as a feedback. I'm aware it's still Alpha, and it is not free from issues like these, as it's not yet the release version. So it wasn't a complaint, but my feedback. 😊 )
Also, one time , after i started the par 1 of the song, so the first part, i had a full 6 minutes of instrumental song, even though i didn't toggle the instrumental option, and there were full lyrics on the lyrics part. It just continue to generate the song and eventally stopped at minte 6. other than that i had also one time a full 5 minte again instrumental generation. besides these 2 occasions, it never happened again.
Heads up, may be an issue on the website. Generations aren't going for some users.
well... it apparently is possible to cause it to get stuck while generating. leaving and coming back doesn't fix it.
You can pick an instant right before it messes up the word and try regenerating from this second only. In the lyrics, remove all lines before the one that’s currently being sung at that instant (that’s how it works for me most of the time)
I spent so many tokens and couldn't... basically it sang my verses and 1st part of the chorus, then cut off about 2 words at the end of the first part, so I did just that... got rid of everything up to that line...and continued... it did finish the line then went into 2nd part of chorus but it sang differently than the 1st part... there's no way I could get it sing the same way... tokens after tokens, generations after generations.. until I gave up... I continued from the begining of the chorus and have a completely different Chorus. So frustrated.
hey guys, new bug today. every ten minutes or so, the website is asking me to accept the terms of service for seemingly no reason. It's just popping up and telling me I can't use the machine until I do it, which is odd. But rest assured, I have now accepted the terms of service like 27 times, and I should be okay
I hope
If it helps, I'm using the website in the latest ai enhanced version of opera
It can only create music. No way to add any singing. Fake and stupid.
the last four tracks in v3 were < 30 seconds, stopping in the middle of the lyrics.https://app.suno.ai/song/439ccb51-4338-4066-b1aa-caea1119df5b https://app.suno.ai/song/1319717d-4d44-4ec2-a671-bb7c1205ec3f https://app.suno.ai/song/02b4a424-d9e9-4954-a30f-55b95cec9800 https://app.suno.ai/song/dc96d507-3652-462c-aa56-fb4cbbe766e8
It can be quite frustrating when Suno generates shorter clips than expected, especially if they stop in the middle of lyrics. To address this, first, make sure you're providing enough lyrics or content for the AI. A short input might lead to the AI running out of material to use, resulting in shorter clips. You can also try adjusting the continuation point from the previous clip, setting it a few seconds earlier, to help with smoother transitions. Additionally, adding detailed information in the "style of music" field can influence the length of the generation. While AI music generation involves some trial and error, these adjustments often improve results.
For now it's not recommended to use Suno. It only generate music with no vocals
Says who?
Thousands who absolutely disagree lol
Right now there's no way to add vocals. At last not in the free version. I used ChatGPT lyrics, custom mode and I added "male singer" to the prompt. Noting. It's not adding lyrics at all. I spent all my free credits. Maybe you must be a paid version to add also vocals, but it's proven that it's not possible to do it in the free version
Vocals are definitely not just a paid feature.
Each one of those links only has 3 lines of lyrics which are rapped. You can hear Suno has reached the end of the last line and you have not provided any further lyrics or instructions to tell it what you want it to do next. Add more lyrics and a [Verse] [Chorus] structure and the song will go longer.
It there a support contact?
hmm ... only half true. It still stops in the middle of words or sentences. I tried adding symbols, empty lines and (Outro) at the end but that didn't help neither. It's probably a problem with the style, so I'd call it a bug (ymmv)
Remember, AI music generation can be unpredictable, and what works for one song may not work for another. Hence, it's recommended to experiment with different styles and inputs to find the best approach for your music creation needs.
as a professional code monkey, I assume you have a stack in which new generated tokens are pushed as they are generated, and the bit sent with the 'stop' token isn't pushed on the stack. This is not a problem when the Engine would generate more than the 2mn anyway, but when there is not enough lyrics/prompt and the output is less than the 2mn it results in this abrupt cut. That's a lot of assumptions here of course, and honestly I don't care about that particular song enough to waste more credits on it, but check https://app.suno.ai/song/fc775c66-b6ce-4d84-b80a-4a2c7da0f32f for the effect I mean.
yeah I know....not exactly what you said but I am aware of this issue...
"Continue from this song" does not work anymore after newest update with playlists; to be more exact: selecting "continue from this song" a song or song part from playlists does not work
so many song generated without voice very frustrated
The website crashes a lot and many times when I want to download the song it thinks I want a continuation so it goes again, so that gets a little frustrating.
The download button is completely gone, I use screenreader and arrowkeys to navigate the web app and while it was always in the dropdown menu, now it's nowhere to be found, also the interface of the website is getting more wild. WHy not leave it as it was? 😦
I've been getting a "Page missing" error all day. Anytime I open a song that somebody shares.
The page actually loads properly and song loads and plays correctly, though.
I'm using Brave.
What screen reader do you use?
Are you able to find the "..." menu on a song? I tested this with NVDA and the "..." button was not registering until it was clicked, but after that I was abele to navigate to Download. The "..." is all the way on the right edge of the song right next to the scroll bar
The only place I see a missing download button is on the explore section. I have to assume that was on purpose. Is that what you're seeing?
When I want to create the 2nd part, i.e. part 3 of a song with the V3, it always cancels the generation.
Yes, only paid V3 getting songs + vocals
I have pro and I am still getting some generations with no vocals with V3 🤷
You have to put the vocals in brackets like this (Yeah) or try "Damn" with ".
I'm using the free version and I never managed to have vocals on my songs. I guess that V2 not supporting vocals
The unofficial Wiki has an article on this that you might benefit from: https://www.suno.wiki/faq/making-music/why-cant-i-hear-my-lyrics/
It's not really working. Since my first try I choose "custom mode", I added lyrics (ChatGPT generated), I put [Verse] before the first verse, and I added "male singer" to the prompt. It's either a bug or maybe V2 / Free version not supporting adding vocals
I've got something on the order of fifty songs that say differently.
Yup, that's what I mean, sorry jumped the gun and forgot to clarify later, it is only missing from the explore section, but that's no good, since I wanted to download other people's works too
NVDA + Google Chrome
Do you mean for the explore or for create section only? I seem to find download button on create, but when I want to download someone elses stuff it won't let me 😄
I'd like to have that download button back too.
The transparency makes the modal windows hard to decipher. I could not read what I agreed to when I clicked the custom lyrics.
Which browser is this with? The transparency is much more subtle / blurred in Firefox and Brave.
I just experienced a bug. I was generating a song and it generated all the way up to 4 minutes. It sounded glitchy. After I refreshed the page, the track was 2 minutes as expected.
Yes.. I had one going to 4.27 and then going back. It happens but it very rare. Devs are aware. Keep reporting things like that. They are valuable!
Sounds good. I won't bother to report the issues I've had doing continuations, as I am sure they are aware. Might be a limitation of their current model. Its very hit or miss, but more miss than hit with continuations.
Luckily the amount of credits for 30 bucks is enough where it is acceptable. And I am still learning tricks to improve my continuations qualities. Like the specific time you choose to start the continuation matters quite a bit.
dunno if this is a bug or not, but its a little annoying, so i was able to use this style for a song, but im trying to continue the song but the style now wont fit into the " Style of Music " and i think its messing up the generation and making the ai change the voice etc.
Can only fit in roughlyy half-ish
@delicate iron Do you think id be possible to boost the character length of the song description and " Style of Music " so its the same, or would that mess with things ?
Continue song broken again. Not pre populating lyrics or song style. Constantly cuts chorus half way through or the last few words. If it doesn't complete shouldn't one be re credited? EDIT I finally managed to create a successful second part that played the chorus through. When continuing the song instead of continuing from the end I chose an earlier point (the original clip had some add libs and repeated chorus so maybe this confused the algorithm. Anyway currently my favourite song, the chorus is soooo catchy.
Tiny UI bug:
I was listening to something on the explore tab, then clicked on the link to the song while it was playing, then moved back to the explore tab. The tab showed the wrong song playing, while the player at the bottom was showing the correct one. The order of the songs in the explore list changed between when I started playing the song and when I opened the tab the second time, so that's probably the reason.
Would like to report a bug. I tried to generate a song with the word "Kalimba", and it rejected it saying that is a known artists name. But a Kalimba is a musical instrument. If I spell it kalimba, lowercase, the song generates.
Actually I take that back, it gets rejected even when spelled lower case.
My paid account after a few days of use became free again and I lost more than 1500 credits. I have complained for the last 5 days and no response. Did the same happen to anyone?
Make sure you are logging into the correct account. It's easy to click on the wrong option 🙂
Why is it adding some random words right at the end of the verse ? V3
https://app.suno.ai/song/f4953f8a-0851-4e05-af67-ba15f8605e9f
Damn my old v1 samples still completely crashing my suno web app
The new poll is nice, however I think an important issue is missing there. At least for non English songs there are a lot of problems with mispronunciation of words. I'd say this is different than missing vocals, they aren't missing, but don't sound right either. With some words this will always happen so I have to simply rephrase the verse to use a different word.
I understand your frustration with the 'Continue Song' feature not working as expected. The issue of Suno repeating the previous verse instead of progressing with the new lyrics is known. This happens because the AI model sometimes doesn't recognize the end of the previous verse. A potential solution is to alter the starting point for the continuation of the song. Try selecting a point just before the beginning of the last verse; this might prompt the model to process the new lyrics correctly. Remember, AI-based tools like Suno can require a bit of trial and error to get the desired outcome, so experimenting with different starting points could yield better results.
When did you subscription start? Is your last bill paid? (<you can see via subscriptions and click on "edit billing details")
You mean hallucinations? They are just like with ChatGPT. Sometimes AI tells lies there, and sings gibberish with Suno. Growing pains.. It's not hard to just continue to the next part before the hallucinations start:
Despite 3 emails over last 4 days I still didn't received a single response from customer help and things are getting worse, there customer service is so pathetic!
What is the problem? Maybe we can help? I know their support team has been ramped up this week so all should be anwered before the end of the week.
Here is the summary of my emails:
Whenever I click move to trash it remove song from trash and whenever I click remove from trash it move song to trash and also when in trash when I click delete it permanently it just stuck there forever despite excellent internet speed and song is never deleted. Also I'm unable to get whole song, it just stuck at loading forever! I even tried a different browser, but it appears that the problem is at AI's end and not at my end!
It's so exasperating!
Yeah that sounds like a bug.. I as a guide have no influence on that sadly enough. But if all goes well you should get a response today 🤞, as I mentioned earlier. If not, just ping me and I will see if I can push it for you. @wary flare
Thank you! I don’t know how I missed this feature, especially as I have thought to myself every day since starting with Suno, “gee, I wish I could just continue from the middle of that track rather than the end.”
I’ve made some amazing tunes so far (10-15k credits in) - but I have needed this.
About to take this thing to a whole new level! Thanks @low cypress !
Yay! I bet your weekend will be a good one now!
Can you, please?
Yes no problem. Just give them today. And if they haven't answered you by the end of this day, just ping me.
Hey, we will look at it
Thanks, eagerly waiting for solution!
Will look at this one too!
We are going to make this one a little more obvious soon
If not fixed, can you DM me the email you subscribed with and someone will look?
Unfortunately some prompts really can make this bug happen more than others.. if you post your style prompt someone might be able to help
for the last 20 minutes or so, all my generations werwe refunded back as they were abot 5 seconds each. It did not generate a single piece. keeps frefunding me the credits.
I don't think it's a bug but I don't see my liked songs I found on explore in my library
Or maybe I'm blind idk
Bug/Suggestion... Enter doesn't create new lines on song description. // Enter doesn't keybind to generate new songs... also anyway to get a partial credit refund for this
I was experimenting with prompting and didn't expect a bunch of generations when I wanted new lines :/
Is anyone else getting odd 'pop' or crackle in the songs being created? Seems to only happen in the V3 alpha when I have 'instrumental' turned on. I've been reporting them as bugs, but I'm surprised how often it's happening (about 75% of the time). It makes the songs unusable.
It might be very genre related. Especially music that has multiple layers of instruments and singing can get muddy... it's not something that they can change overnight. This is just growing pains.
Try minimalistic music, and you will see the difference.. like jazz, hip hop, folk music, unplugged etc
That could be. Here's an example of one of the tracks. They're throughout, but most noticeable right after the 1:00 mark. Just wondering if anyone has found a way to combat them. I understand it's just growing pains, just hoping maybe someone found a workaround. Overall, it's hard to call this tech anything but amazing.
Yeah.. electronic music has so much SFX that it is hard for AI sometimes to keep the audio clean.. this is normal for now. The suno team strifes for better quality but it's not easy to learn the AI to pick it apart and remove the SFX and then revert it back.
But in the future anything is possible
I understand. Thank so much for taking the time to respond @low cypress . Have a good day!
Yw.. have a great weekend 🙂 @honest harness
I work on an ipad and since yesterday (i think) it crashes a lot when you leave the site ( for copy the continuation text and want to past it in Suno) it’s annoying because you can start over
sometimes its really frustrating second part of my song the verse starts at the end and goes on
Hello, thank you so much for creating this. I keep running into some troubles when it says "Credits refunded" I have found that I am not getting the credit back. I have the pro plan and I should have reached out first before I kept going. I now have 1,255 credits and I notice this problem has happened a lot of times. Another concern is that it will give me a 10 sec clip with a quick sound and the rest of it is dead air. In this situation it doesnt say credit refunded but acts as if its a quality clip that is credit worth. Is there any help I can get for these situations?
V3 hasn't been giving me anywhere near 2 minutes of generated music. It's usually less that one minute. I'm assuming this is the same for everyone. Also noticing that the audio quality continually falls apart after the initial generation. Sometimes really bad with a ton of white noise.
It only takes money afterwards if the part is created. So not in advance. Not seeing money returning is therefor not happening, since it never took anything. Test it yourself. Create a new part, and watch your credits. Only when they are longer than 10 seconds and rendering, it will take your credit.
@analog ibex to achieve better audio quality with the current version (v2 and V3 alpha), it's recommended to try generating music in genres that are less complex and use minimalistic effects, such as unplugged styles or hip-hop, which tend to produce clearer results.
@analog ibex Suno's V3 version can struggle with utilizing the full two-minute length for music that includes lyrics, often resulting in shorter clips. To potentially increase the length, try providing more detailed information in the "style of music" field when generating songs, different styles of music, more lyrics or making pure instrumentals. There are plenty of people who manage to make 2 minutes the first part. But this is still alpha, and AI can be fickle.. Thank you for your feedback!
Thanks, but I'm not interested in acoustic or hip-hop. Also not interested in instrumentals. I need vocal tracks with full arrangements. Looks like I just have to wait and clean up the files as much as possible offline until everything is working for all use cases and not just some.
I really get where you're coming from, wanting to use Suno for more complex vocal tracks with full arrangements. It's frustrating when the technology doesn't yet meet our needs. Remember, Suno V3 Alpha is groundbreaking but still in development, so its ability to handle complex music genres with high-quality vocal tracks is a work in progress. It's like expecting a young sapling to provide shade like a fully grown tree—it's just not there yet. Stay tuned, though, as Suno is continuously improving, and what you're looking for might be available in future updates.
I'm pretty confident it will get there. There's nothing even close to Suno at the moment. I'd actually be willing to pay more for an unlimited plan since I need a ton of good music for a project I'm developing. I only use about 10% of what I generate because the bar is pretty high. Do you know if anything like that might eventually become available or if stems/separate track downloads will be available at some point?
These features are in high demand, they are being seriously considered as part of Suno's future updates. Anything is possible in the future, for sure!
help me:( command not work
looks like you longer you use it it gets worse tbh. i made 3 good songs and tried 3 new ones but wasted credits because didnt executed well enough or to short. But i really think it will get better ^^
when i want to generate a song idea in prompt maker then the genre &/ the song subject don´t gets displayed in chat gpt 4 prompt maker then rest is fine but the essential thing style and Thematic don´t show up
Ok
Welcome! For help, check out Suno Help
Welcome! See Guides and Resources for help using Chirp and Bark. Email us at support@suno.ai if you have questions, comments or feedback.
v3 sounds bad now, its not just me its also with other people, v2 sounds much better now, idk whats going on
Both v2 and v3 seem to stop right on the last word, cutting it off. There should be an option to trail off and wrap up (but not automatic since I might want to continue the song)
Nor did I received any email neither is my problem solved!
Not my experience. v3 sounds a bit less variative to me, and the sound may sound “thin” but that’s actually because there’s less noise. And keep in mind it’s alpha. Here’s an example of v3, I doubt you can say this sounds “bad”: https://generated.bandcamp.com/track/believe-in-me
How much lyrics do you input? Sometimes it happens when you just enter say verse 1. IMHO that’s a good thing because then you can decide where you want to go from there (either chorus or next verse or even ending), you can even go in several different ways and then combine them in the needed order in a DAW.
I've send fourth email as well but still no response!
Who asked? That was 4 days ago.
You can literally see the message I was replying to from earlier today or yesterday
You promote yourself on a bug report channel with a 4 day old song to someone complaining about v3 now.
Please check the website app ; when doing a continuation all the clips disappear, only to be found at the library, not at the create section anymore ( this is on an ipad in a safari browser)
Actually I reeeeaaaallly like what Suno V3 did here out of a few words, especially the guitar sound on this track.
It was re-worked in a daw to get more "room" and brilliance into the mix, but at 1:26 you will notice a sudden lack in sound: the guitar degraded very fast and ended in mostly clinking sounds.
I tried to get rid off it using a FFT-Filter, but it had to become very broad, so wanted frequencies also disappeared.
This "degrading" still happens in long tracks with distorted guitar sounds, but not always.
https://www.youtube.com/watch?v=SADHJi3HZ-s
Shoegaze with a Funk twist - see the album.
lol, I’m not promoting by showing an example. I don’t see how its being four days old is any relevant because I don’t think any given versioned model changes over time. But in case it is (and to piss you off just a tad more), here’s one from yesterday.
@low cypress Still no response or solution even after I send a fourth email!
It stuck there forever!
I am gonna try to figure out why it is not answered yet. Although it is the weekend, so I am not expecting action before Monday. 😬
That is all I can do for now...
I need my song before 21st March, hopefully it will be solved before that!
To avoid songs abruptly stopping on the last word in both V2 and V3, try adjusting the continuation point a few seconds before the end of the previous clip and ensure you provide enough lyrics or content for the AI to work with.
The degrading sound in long tracks with distorted guitar sounds, like sudden pops or crackles, is known to occur in the V3 alpha, particularly when the 'instrumental' mode is on. This issue tends to be genre-specific and more prevalent in music with multiple layers of instruments and vocals, indicating it's a part of Suno's ongoing development.
Im looking Forward for the improvements. Its so nice to have his own music to listen to.
For longer songs, I suggest picking a part like the end of the chorus and continuing from there several times. This way, you can have separate “songs” for say verse 2, solo/bridge, ending, and then combine them together in a DAW. Having a verse and a chorus in place provides solid foundation to ensure the consistency of what goes on afterwards.
That said, I think degrading only happens if you’re dead set on following the same patterns. Sometimes the model will just end your song and start a new one, and this case the “degrading counter” is “reset.” I have multiple songs that ended up being split into two songs.
I made a song, the song is very good, and the song is also very good, but the last sentence is missing two words. What should I do to make the lyrics of this song sing completely without affecting the tune of the original song? Just sing the lyrics that you sing less, and the rest will remain the same. Please tell me, thank you.
We are continuing to look into it - I saw your message yesterday. Can you DM me a link to the song you are trying to get full song from?
I've sent DM!
hey, is it only me or download is not working?
If you’re a paying subscriber, use v3 and continue from just before the sentence you want to correct.
I'm getting 2 minutes about 50% of the time, get occasional noise degeneration but overall the sound is cleaner and more stable than V2
I'm getting the cutoff bug when the AI has enough content to fill almost the entire two minutes, and when there is no continuation (first run). Adjusting the continuation at the start to fix a bug at the end is also weird
note that this is on the web interface
Hi guys, im having trouble getting the create tab in the web app to load anything at all. ive tried different browsers and devices. any idea what im doing wrong?
The Suno team is actively aware of this issue and is working on refining the balance to achieve the best outcomes. Adjusting the continuation point, as you mentioned, might seem like an unusual workaround, but it's part of experimenting to achieve better results in this evolving technology.
Maybe there is a problem with your internet provider (network issues from route A to B)? This sometimes happens. If you have a VPN it would solve it.. It is ussually temporarty. I can personally experience no problems myself atm.
The explore tab is working, its only the create tab and library tab that just load eternally. ive tried it on my phone, which is a different internet connection, and that didnt work either.
V3 Alpha seems to have a persistent problem of ignoring new verses. When I click "Continue from this song," and then type in the next verse, V3 just ignores the new verse and instead repeats the lyrics from the segment it continued from (albeit a bit garbled). Also, V3 seems to have considerable trouble sticking with a melody/cadence/rhythm once I begin "Continue from this song." I have no idea how many credits I've wasted tonight because segments it generated were completely different melodies, rhythms, etc.
Compared to V2, so far my experience with V3 Alpha is that it scarcely rivals what I presume was a V1. With V2, I can usually nail a song within 3 to 4 generations. So far with V3, it either never gets it at all or needs 20 to 30 generations minimum to even stay consistent.
Still a VPN would be great to test it.. Since I know not all elements are loaded from the same server. And since you are the only one who has this atm, it is probably something on your end of the line so to speak.
Ok, ill try setting up a vpn.
Not sure if this is helpful. i dont currently have a vpn, and dont really feel like paying for one. do you have any recomendations?
this is a guess, but are you logging in via discord? if so, you may need to verify email via Discord (this may be causing the issue on our end). https://support.discord.com/hc/en-us/articles/213219267-Resending-Verification-Email
Here's an example of the cutoff
https://app.suno.ai/song/b87f94f6-0c85-4331-a1bf-4fac69cc9a4b
Happy birthday to you,
Happy birthday to you,
Yuh a year older, nuh look nuh frown,
We a come celebrate, mek de vibes go round.
Blow out di candles, mek a wish come true,
Cut di cake, we a share it wid you.
Sun a shine, music a play,
Happy birthday to you, every single day!
This is not a continuation, and it cuts off right at "day"
which is perfectly normal/ You should continue.
oh, so to end the song, I need to continue it?
yes, continue, and in the lyric field put something like [end] or [fade] and there's a good chance you will get an ending
Thank you!
When using v3 to make songs on the jump style genre, I get little to no song progression. It's like it isn't recognizing the usual [verse] [pe-chorus] [chorus] tags.
I used nearly 1000 credits yesterday and couldn't get anything worthwhile.
You're a genius! I got it working in chrome, but still crashes in safari mac and ios.
actually its working across the board. just needed to add my email to discord, thanks guys
Sorry, we need to handle this better on our end... glad it works
the twitter one was a few days ago and the lemon one was yesterday https://app.suno.ai/song/3aa0f433-fe31-4ffc-8284-1fdbf1e61989 https://app.suno.ai/song/1efa2a2e-6b12-44c6-b176-cbcef399c9f4
indie rock, quirky, male singer song. Listen and make your own with Suno.
I really appreciate how considerate your responses are and how you understand people's frustration, whilst being realistic about what we can expect from Suno at the moment. Thank you ❤️
buy more credit is no more available???
yes it is?
no not present
What do you see then?
Oh you need a subscription to get the option to buy extra credits.. 😬 @dim dagger
ok I c and I understand now on thank
another point : how get clear voice stable when generate part 3 , it's look like it loose tense for the last part
[Outro] never give an nice ending always cut
Trial and Error: AI music generation can be unpredictable. If you're not satisfied with the result, try adjusting your inputs and experiment with different styles or lyrical content.
To end a song in Suno, use metatags like [Big Finish], [Fade Out], or simply add "end" in the "style of music" box without quotes. These triggers can effectively signal the song's conclusion.
ok thank again
If it is a continuation, try adding "finale" to the style. It doesn't always end it, but it often adds more emotion to the ending.
i tried to continue a song in v2, and the original song, broke and the Ai would repeat the same word, (a single word) till it reached the new prompt and continued normaly? did i do something wrong?
The issue you're facing, where the AI repeats a single word before continuing normally, could be due to the bot losing its place in the lyrics. To resolve this, try adjusting how you continue the lyrics, like picking up with the next word or adding a line that was already sung as a lead-in. But for that you need to be able to continue from anywhere which is a new feature that comes with paying subscriptions.
okay, i guess its a v2 issue
No, it's something that can happen with V3 too.. New tech so growing pains in general.. Just like ChatGPT lies, AI music can hallucinate or get lost where it should continue.
what if i include the lyrics from the previous prompt(original prompt) and include the new lyrics i want to add in the continuation?
There is no certainty. You could try. Still a lot of trail and error.
alr. thank you\
I know you're past this now, but proton vpn has a free tier.
Thanks, I’ll check it out
Seems that V3 is barely capable of producing anything over 50 seconds for some reason.
I got it to produce up to 2min songs so far... It's true that continuation seems to Las only 1 more minute, but you can keep on like that forever.
By design. V3 can do a 2:00 initial generation, then it's always only up to 1:00 afterwards
It helps to get a full song and continue from that, rather than JUST continuing from the 1:00 clip
OK guys, your AI is fantastic and the v3 is baffling. There is just a single feature that is really missing I think : the ability to regenerate a part of a song. We can only continue from a position, but there are many reasons why we would want to specify an end point where the song would continue from.
You mean generate backward? Like regenerate the intro?
I think François means paint in... Read this: Suno is currently researching a feature called "paint in," which would allow users to select and regenerate specific parts of a song, like a bridge, chorus, or verse, without changing the rest of the track.
Then yeah inpainting as the visual generative AI people call it
Music has a temporal aspect to it images don't so
and you're right, Suno does need that
I give you an example very common: imagine that you got a perfect generation on your 2min, but the first verse is not good. You'll never get Suno to randomly give the same result, and if you use the continuation tool starting from 10s after the start, you'll likely get something completely different
Yes, that's likely what I mean
Thanks
It's coming
Ah damn, I'm on the wrong channel anyway. Here, I wanted to mention that I got a few generations where I got no lyrics at all. Probably a kind of bug. You probably already know about it?
Thanks! Why is that faq not in the faq you access when you are in Suno?
it's communty wiki created by the community
Suno is still a young company. So many things still needs to be worked out including better faq 😄
That's why we as a community help them. I am for instance a volunteer guide 😉
Any idea about the number of users they have, just to get an idea about how far they are in their development?
I can see the discord which is 285 k users. And I suggest for more info the #📢┃announcements to see the progress so far.
I'm thrilled by what it might announce for the future. It's a super tool for small artists, and it could help improving the quality of songs world wide. I'm looking forward to see this 🙂
i think the moderation tool chain is buggin out, it seems to look at words instead of context and flag for moderation. a quick work around could be to implement a dispute solution for moderated prompts... at least this way the artist could explain why the prompt exists, and may be allowed after that
Hey y’all any devs awake? Suno is down
@tame latch ^
looking
It's cause I literally just bought the Pro subscription... woe is me haha
should be back
it is
seems to still be thinking about stuff I submit right before it went down, should I kill those jobs?
how do you kill jobs...?
be a Politician
I just deleted them, they appeared to be stuck in generation mode
Everytime I generate a "Japanese Pop" song on the site it keeps failing
Strange, maybe try "j-pop" or some other way of expressing it
i tried "J-Pop" and it only slightly decreased the failure rate
This would be the coolest Feature 👍
hi, I would like to know why it always generate noisy voice on part 3 or 4 all the time, it's very frustrating, my song ha intro-verse-chorus-verse-bridge-chorus-big Finish, is it too long for the IA ?
is there a way like generating IA image to insert negatives prompts ?
Are you using v3? It is much better about the sound degrading on continue than v2 used to be
yes I'm always using v3
I love getting a pretty bad ass song to only hear half of it on v3 t(-.-t)
i continued a part that end with the word "vulnerability" and every continued part starts with "strengthen" or any other word yet those words are nowhere in the song
The System sometimes hallucinates like every LLM typically next Token Prediction. What helps for me is using the Last line from the Former Part in the new Part
wha-?
The orchestra music generates a copyrighted melody. It has appeared in a few of my tracks. Is there any way to remove it from the AI? I don't want to get flagged for copyright violation. It appears to only be generated in Epic Orchestra tracks as far as I can tell.
I personally don't think so. Can you post the music so I can check your statement?
Yeah in a bit, working atm. I will load the track and the copyright notice i got from Youtube. @low cypress
Thank you.. please an url of the original creation in Suno.
It was done in a batch of 50 instrumentals; I am not sure how to find that specific one. It was a few days ago. I dont keep the suno copies after I load them on my editor. What I can do is the next time I am doing that kind of music, I will save all the suno copies so if it appears again, I will have the suno copy.
Yeah to figure out what is wrong, we need to go to the root. Aka the files you worked with. If you still have the full song in your library that would help.
I wasnt aware there was a problem till after the fact, or I would of kept all the suno copies. I will be ready the next time it pops up.
Out of 50 tracks it popped up 1 time. And a week later is popped up again 1 time in a batch of 50. Its the exact same melody.
oh i think ik what you man now. Skip a line on the continued part before continuing the chorus
ok nvm that was just luck. Apparently it struggles with continued lyrics after it saying "vulnerability".
Not skipping, repeat the Last line in the continuation without Meta Tags like Verse it chorus example:
bla bla vulnerability
[Verse]
new Text goes on…
Or when the very last word gets cut off...
I'd like a feature just to add a few more seconds to the song and finish it.
Aren't you aware of the continue option? If you do it right, you can achieve just that.
I am absolutely aware of the continue option, but having the song end too soon is my #1 frustration.
Some songs only need 1-5 seconds to finish them, but that is considered an error in Suno and credits are refunded.
Yeah well if you want totally control, then I have to disappoint you.
Yeah I get clips of that too.. but I consider it a bonus. I got my end but didn't have to pay for it. You can just use it to make that full song.
But it is not predictable. Sometimes I was just a few seconds and it goes off on a wild 2 minute hallucination. Sometimes I want that full 2 minutes and it cuts off at 37s in the middle of a verse. It is practically almost always the opposite of what I want.
I will explain you I do it which works in 9 out of 10 for me...
So let's say I have a song that is already fading (that works the best), then I continue from that part, remove my lyrics (leave it fully empty), remove the "style of music" and just put in "end" without quotes. That works perfect... Also if you have a clip that fades out and comes back, just set the time where it fades out and do what I mentioned.
If the song is still going, this works too.. But there it's 9 out of 10 with me.
But keep in mind, a lot of things are not predictable. You could wait and see how development goes, but you want to be and stay ahead of the curve. At least that is how I think 🙂
My more common problem is the first generation having the song cut off on the last word or mid verse and also well before 2m. I know I can fix that by repeating the missing part and continuing, but you end up with a complete song with lyrics that are all jumbled up. If I only wanted the mp3 it wouldn't matter. But I'm often sharing them with people where having the proper lyrics that match the song would be important. It requires more post processing to re-add the proper lyrics and make a youtube video.
Fully agree.. one of my suggestions is being able to edit the lyrics afterwards and maybe even regenerate some images...
Thanks @low cypress, you are a support champion.
Hi, I've found that when I make a Full Song sometimes the final parts are cut and therefore not real song endings. Do you have any tips for developing a complete song correctly?
You can still continue and thus ending it.. Even with full songs you can continue
thanks. how do you generate a final part? do you have any advice?
do I need to add some meta tags?
@orchid owl I explained it today to someone else. See: #🪲┃bugs-web message
When creating Chinese songs, Mandarin and Cantonese are often mixed together. These are two distinct pronunciations in Chinese, so please do not mix the two pronunciations into "Chinese" pronunciation. Instead, use "Mandarin" and "Cantonese". In addition, the pronunciation of "Cantonese" is quite inaccurate.
Alternatively, temporarily retrieve the AI for Cantonese, as the current AI pronunciation for Chinese is messed up.
Completely unusable.
Chinese Mandarin and Chinese Cantonese are two completely different Chinese languages. Do not train them together.
Currently, it is a waste of credits. I paid for it, but every time I use it, hundreds of credits are wasted because two Chinese languages are mixed together, and the song cannot be used.
Try to explicityly name the language both in the style box and in the lyrics part between brackets [Chinese (Mandarin)]
The style description “Freestyle” is blocked because it is an artist’s name!! Could you please unblock that?? I feel like that’s very limiting because it doesn’t have to do with the artist at all, it’s a way of creating music and a term that would be useful for user’s to explore more free-form music generations!!
👁️🗨️👄👁️🗨️
😭🙏
They cannot.. copyright
That’s so fucked up!!! 😭😭😭
"Couldn't generate that.
Tags contained artist name: emin"
Emin = E-minor key is banned?
That is nothing they cannot change anytime soon. World wide copyright laws are to blame.. not Suno
It's also about creating and generating your own songs, not copycatting somebody else
I agree!! The laws are fucked up!! We shouldn’t be restricted access to a word like “freestyle”.
That’s not what I am trying to do. I want to get the singers to have more freedom in their singing.
I thought they already whitelisted that word freestyle.. I guess not.. because general words they can whitelist
🤔
Could you ask them plzz?? Maybe they forgot to whitelist it again!! 😮
I will 😄
Thanks!! 🥳✨💖🤩
Continuing a song without lyrics doesnt work. Especially edm, coins always refunded and I insert the same genres I listed into the previous portion
I think you are doing it wrong then? If I flip it to instrumental, or write [instrumental] in lyrics, and give in my music genre, for this example, EDM.. it gives mostly 2 minutes of EDM
Does this refer to first continuation also? or just for later ones? And what input do you put in the lyrics/custom prompt input for the continuation?
The issue with noise in V3 Alpha, especially when creating electronic music, is known. This genre often involves complex sound effects (SFX) that are challenging for the AI to handle cleanly. You might find better results with genres that have simpler compositions, like minimalistic music or unplugged styles.
Thank you. I hope V3 can sort it out.
@dusk thorn @low cypress I have instrumental on, same genres listed, same issues. This refers only for continuations
It might be because your song processing reached a certain step so it cant generate more... i usually - even for instrumentals - use the lyrics input to make custom processing like [Drop] etc. on continuations - not letting it randomly continue....
I did it with lyrics and just put "(Instrumental)" still no success
or when much fails: using [Break] which sometimes resets processing
Use [] for Meta Tags in the lyrics input Im not sure if (Instrumental) works....
Try [Instrumental] or try something totally different
Okay. will try.
[Break] and then give it real instrructions how to continue.... maybe use [Main Part] or [Climax] just to see if it changes something....
Nope didnt work with either.
even with [Break] as first input?
ill try again 1 sec
maybe internally your song has been marked somehow as finished.... i had some same experiences like you too, rarely, but after 10-15 tries i got something that let me continue regularly
how many tried did you do now?
atleast 8x
your initial result, how long was it?
Try adding “symphonic” , “orchestral” , “virtuoso” . Terms related to more classical instrumental music guide it towards more complex compositions and clearer acoustics. I don’t think “clean” will help, as that’s too specific for current models. Employ more widely used terminology related to genres of music and styles, not so much the actual components of the song or technical aspects of it.
i know i had to get once like 15 tries to get a result that was not a payback of credits or just like 2 seconds crash or something. But thatg was when i did a more obscure experiment....
I will definitely try this, thank you
Virtuoso specifically is INSANEE!!! Glad I could help!! 🥰✨💖
My last 15-20 generations have had no lyrics at all, despite me having not very complex lyrics in the lyrics box V3
I'm noticing considerably more noise in the last few days. I've been working on the same song for a week, so I think I'm able to judge. I wonder if they updated the model in the last few days?
feel like i'm throwing away a ton of credits to generated stuff that's going directly in the trash. Something's up with V3 :/
same thing happening with v2
If it was that easy, it was already done 😄
No, they haven't changed the model. However, the process of generating music with Suno's AI involves a complex interplay of different elements. Sometimes a lyric might not mesh well with the AI, requiring adjustments. There are instances where a genre might not produce the expected outcome, or the combination of lyrics and genre doesn't work as intended. Metatags are also pivotal; using the wrong ones or not using enough can lead to unsatisfactory results. Remember, AI music generation, particularly in platforms like Suno, is still in a phase of continuous development and learning. Prompting effectively is a skill that can significantly enhance your experience with Suno, as it guides the AI more precisely to create music that aligns with your vision.
@umbral falcon the above answer probably also applies to you. 😉
I haven’t tried this, so I don’t know if it works, but maybe if you set the song generation to the instrumental setting you may get better results.
@dusk thorn the inital song was 49 secs ending on buildup https://app.suno.ai/song/9c4f86e2-2f99-4b7b-970d-e0a4e1e0b6ff
Dubstep, Future Bass, Drum & Bass, Hardstyle, and Industrial Techno song. Listen and make your own with Suno.
The AI may be fighting between leaving space for singing and including instruments, so that may he part of the reason why the noise shows up. It may be the beginnings of an instrument coming in but then getting drowned out by the AI wanting to leave space for a voice. Just guessing! 💖
I feel like I'm getting ripped off and throwing tons of credits away when the clip cuts out after 3 lines of text and in the middle of a word. Why does it often not generate more than 20seconds? I would prefer it to keep going and cut it myself than try to patch in words that haven't finished from one clip to another. A clip max length makes sense but I'm being charged for clips that only last 12 seconds with 8 seconds worth of silence.
I understand the frustration with short clips in V3 Alpha. This version, still in its alpha stage, is undergoing growth and learning. For a more stable experience, you might want to try using V2, which is currently more reliable. However, it's important to have realistic expectations with Suno, as it represents the cutting edge of AI music generation and is still in the process of being refined. Plus, good prompting is a skill that needs to be learned as well.
That sounds like you hit a very narrow result here
