#📚┃suno-school
1 messages · Page 90 of 1
Im assuming i should stop creating right now to save credits and wait until later today then?! haha omg
Will using the V4 Model cost us more credits??
remaster seems to cost 2 per 10 credits 😮
Should be the same price, we will know the duets when we get the announcement
damn didnt catch that! gotta say worth it!!! AND got the BEST Persona Out of it. absolutely next level
Remaster will likely be (X) Free amount, probably 200?
only 1 free
Has anyone tried an older persona on the v4 yet? Does it sound any better?
but there are 5 for mobile....whatever that means
So these voice tags, can they be added into the lyrics to tell the AI that on this sentence, I want you (as an example) to sing lower? https://howtopromptsuno.com/making-music/voice-tags
can anyone know if there's a term for a key pitch in a song when it raises the key or pitch of the song for dramatic emphasis? does it raise a half key? full?
I was just gonna go out for some groceries.... I guess I'll starve to death now 😵
Hello, will we ever have API integrations?
The announcement post says personas and covers are “supercharged” in V4. Can we get more info on what this actually means? Do these features actually work properly now unlike before? This is pretty vague terminology.
All of the improvements that v4 offers, but with the added flexibility of changing styles 👀
Whole AI music is one big vague thing to be honest 🙂 You just need to embrace it
There is a issue for the song time ? Many songs is cuted at 3:12 😮
Is there a reason why I can't cover with v4?
I would love to know if they fixed the persona issues with v4, but nothing's said about it...
Well it determines whether or not I’m gonna resubscribe. Like many, I had a lot of issues with covers & personas working properly so I mostly wanted to understand how they were improved. For example, half of the time I would do a cover it would just output the same audio with a slight key change. If that issue is fixed then it should be mentioned.
try refreshing the page?
.
The new lyrics model is less restricted
Not a big deal -- I just realized it only impacts the lyrics that are autogenerated -- the automod is just as aggressive as before
That may be subjective. Feel free to ask other people, check v4 songs or... try by yourself
I’m gonna wait and see what other people say. I can’t resubscribe with such vague info like “supercharged” lol
2 things seem to repeat with v4 - better sound quality and more broad sound
Ya, thins sound so much different now, some are much better
If you're using a persona for those, it sounds like that particular persona is using a 48s clip of the original song the persona is based on -- that's what it uses to ensure that the follow clips sound similar.
I still have free persona songs availiable, I'll test on one of the ones I'm working on right now
Can't cover a song made with 3.5? I can only remaster it?
but what is "persona"
Hi! So I don't know but like I had songs I used with Suno but I can't see them anymore?
Is it jjust on the web browser r what
So songs with certain types of drums are turned into 8-bit drums. Anything heavy
I have a pro subscription and can’t make covers or songs using V4… I change it and it keeps going back to 3.5
a persona is a style you save from a generated song
Basically think of it like an artist/song profile
aw.. they removed my free credits on personas LMAO
looks like they fixed the lyrical issues with personas. It's not generating the same lyrics anymore edit: Nope, it STILL generates the wrogn lyrics
its spinning when generating a song
but why they delete our free credits?
can take a bit
I can't select v4 for covers. Only v3.5
but its been like 5 mins
Okay so here my at small review at the v4. It is good for "normal" music. But when there's some techno in it, the ai have some difficulties. A lot of reverb mostly. But for "simple" music it's good
Any screen reader users here? I can't switch models. Does this mean i'm stuck on V3.5 until they make it default? Or do you screen reader users found a way😞
And? think about how many people are currently hitting the servers
use shift-tab until it goes to it
Here's some exemples.
https://suno.com/song/d3bb2d67-6729-43ef-9fba-82195472c732
eurobeat, fast-paced with eurobeat riffs and eurobeat intro, supersaw, j-pop, triplets, 2 singers, electric guitar song. Listen and make your own with Suno.
80's, pop, catchy, trumpets, synth, electric guitar bass song. Listen and make your own with Suno.
Disney song, Evil song, Romanian accent, Musical, orchestral, chorus, violin, cello, trumpet song. Listen and make your own with Suno.
same here, i'm not sure if i'm using v3.5 still or v4
it says v3.5 still, but at the top i'm using v4
When I select v4 it generates a song about this is a fake app and i should go to suno.com to download the ios or android app....
https://suno.com/song/35c23a70-bd2b-4bac-9669-320bb752984d
dark trap, experimental hip-hop, trippy song. Listen and make your own with Suno.
Lmao what, that’s weird
Is here an Admin i can Talk with?
Try changing models to like 3/3.5 then back to 4. Could be a visual bug with default set to 4
it's not defaulted to 4 though
If you need help with your account or something the community, reach out to support@suno.com
but i see them fine at the main menu
Doesn't work, it goes to upload audio... Are you a screen reader user?
hit tab again, but not shift-tab
Please contact support@suno.com
Personas that were created on V3.5 get an error when trying to create songs in V4.
are you actually on the IOS app?
Its a Thing of billing
V4 is out! 👀
that's intertesting, I made a v4 cover with a v3.5 persona without an issue (other than the v3.5 issues, of course)
In that case, definitely email billing@suno.com because no one here would be able to help you with that
im on it like a fat kid on a smarty
Can you please tell how much time does it take to revieve a reply
Thx
It worked now. I had to create a song in V4 then go back and select the Persona to create another... Oh well maybe it was a gremlin.
Typically a few business days (EST time zone)
Ok ty!
So you have to user mobile now?
that could be it because you'vd never used v4 yet. I remasters a song first
I hope that it can help to see where's the v4 lack in
yes it's releaseddd
I can't still select it. Screen reader (NVDA) moves to an unnamed button and when I press it, it doesn't switch models. Guess I have to wait longer than most people... unless there is a solution. I know @clever crescent uses a screen reader
hey do you mind sharing the channel link where the pros rub elbows? Thanks 🤔
I do not, but that does get to the button. I'm lookign for how to select it within the keyboard
nope lol
Does anyone know if personas are based of the model? Like is a persona from a 3.5 song gunna be the quality of vocals from 3.5?
I might be bias -- but there's lots of creative innovators right here in this server, except they don't pretend they understand everything -- they've admitted that we're all learning in the same way, and want to share what they've found. 💪
@frigid orbit has been sharing some of their observations in #1252267444773326848 to name a more recent example, but there's a lot of great community resources 🙂
persona is simply afingerprint, it has nothing to do with the generation
Can you create a persona from a song that used a sample input yet?
V4 remaster is great for vocals but the instrumentals are watered down a bit 🤔
Doesn't look like i can remaster with a persona. So I lose the vocal style when remastering. Maybe just need to do more atttempts though
V4 is great, but am getting a faint repetitive sound like a distant burglar alarm in my remasters. Thanks for all your efforts to improve SUNO!!
https://suno.com/song/760aefe7-9fab-4663-abd8-e464f3a544ad <the persona issues are still very present here
dark grunge clean female vocals, guzheng song. Listen and make your own with Suno.
Wow the extend feature in v4 is weird.
Thanks anyway, I feel your pain, sometimes making music can get "painful" but hang in there, have a good one 🙃
you2
anyone here having issues with buying extra credits?
Have zero credits left for v4. Doh. At least it refills for me sometime tonight.
Is it me or there seems to be a lot of background clanking in the remastered versions? Not sure how to describe it...
They use some non-standard method for pulling that menu, no shortcuts I have found work
Yes I'm getting a clanking ticking type noise too
Same here. There's a really wierd sould in the backgound that makes the tracks unusable. Here's an example. Very noticable from 00:40:
monk choir, church, techno, up-beat, male vocals song. Listen and make your own with Suno.
This really should be reported, I'm sure they can tune it
Is it just me, or do my metal intros in Remasters sound weaker than before, and the vocalist starts out different, eventually 'morphing' into the original?
can you link the original please?
Yeah, here's the original (made in v3.5):
monk choir, church, techno, up-beat, male vocals song. Listen and make your own with Suno.
I'm getting that as well in the remasters
I am going to suno.com directly and it's not letting me in it saying on every song for the lyrics that is the fake app or something like that for me to help me out?
Hi there.. V4 remastering works for someone? I'll try and try and nothing
I noticed regarding the creation of instrumental music that on rhythmic genres (drum and bass, breakcore, electro house sometimes) there are hi hat or cymbals noises or something very fast in the background.
Whoops speech to text butchered that
if you remaster it again, does it do it a 2nd time?
I can select it, i press on v3.5, go down to the end of the page and just press on the model I need.
Yeah, I have tried several remasters. All with the same sound issue
What if you reuse it and generate it anew?
Have some people tried to add additional voices to Suno? I also tested adding notes to try a French Christmas song, and it's totally terrible. It sounds like rap, and even when amplifying the sound well.
... > Report > Bug > Poor Audio Quality & thumbs down is remarkably powerful in regards to user-feedback in these early days 👀
Even with a "reuse prompt" it's still noticeable in the background
does anyone know what quality the waves we are downloading are in V.4
Is there any other contact besides billing to resolve a subscription issue? I haven't heard anything in nearly 24 hours.
Have anyone of you tried making a persona out of a 10 minute track? It seems to not want to do it and keeps erroring out
Hi, I am impressed with the V4 masters, but I cannot seem to change the model in create box from 3.5 to V4. I am using a screenreader if it helps. Any ideas?
I usually did that and it worked, but this time it doesn't seem to change anything and there's no choice to change the model
I forgot to tag you when you asked last night, but have you tried to crop a 9m segment of that song & use that for a persona? #📚┃suno-school message
It seems like personas use anywhere from 12-60s of the original song, so cropping out a minute to let it fit into the persona window shouldn't impact the resulting quality of the persona, I'd imagine
It worked for me, One thing tho, It shill shows v3.5, But I think this is just the bug with model names. Song do seem difrent.
lol song in question has 2 minutes of buildup alone
;-;
Has anyone managed to create a full song lasting more than 4 minutes? And if so, how can it be done?
They're the only people to contact for billing issues -- active during american EST business hours, and seem to respond to emails in order of when they were received -- it might take a couple business days 🙏
in 4.0 some weird noise comes in , anybody know when it will be fixed?
Extend, then "get whole song" -- 4m is the limit for initial generations -- #1270928841614823546 message (a video tutorial on how to extend)
so no remastering v1 songs?
Thx
Has Suno V4 increased the quality of the songs wavs that we download.?
If you're giving it a 10m wide sheet, and it's only able to "cookie cutter" out 1m max to use, shaving off 1m of the intro likely isn't going to impact the persona, seeing as how the persona doesn't use the full 10m, only whatever small clip it chooses to focus on -- you can get more specific results with a persona by priming the clip & giving it less to choose from, so you have a better idea of what the resulting persona will sound like
Are yyou using a Audio interface, if you are some of them will give you static noise.
No, it's a whole v4 thing 😅
Not static , Its almost like an arpegiator
When I do remastering It just gives me 2 seckond silent clip, and refunds credits. So I guess you people are lucky with your remasters.
Just tried extracting a 1:48 long segment and making persona out of that didn't work either
Seems like the servers might be SLAMMED atm
Did you download it as wav.
If you're not able to make a persona from the song using even a small clip, then it sounds like the clip might be the issue -- is there any chance the clip originated as an audio upload that contained vocals?
No it isn't an upload
Yeah
... > Report > Bug > Poor Audio Quality -- a couple different names for that particular v4 artefact atm, but reporting it is the best way to help them train it out of the next update 🙏
Although one thing I am seeing is because there are so many extension when I select range to crop it doesn't always pickup all the lyrics and tags in red
suno v4 its working?
Alright
to me its not generating
Maybe it doesn't know what lyrics and tags are for what parts of song so it can't make a persona?
idk
What quality was the audio.
If you want to dm me the link, I can see if I can figure out what the hiccup is?
The whole song is very nice , It's just that one thing that comes in
https://suno.com/song/63bf684d-1458-4ba1-a01d-ae7eabbc0820
Some one ordered Devil? XD Why it has this strange intro XD for a whole minute+
Electronic, Beat, Circus, Dark, Bass, Scary, male vocals, RnB song. Listen and make your own with Suno.
Upload the audio.
@limber pike EVen the Sample songs suno officially has, they have it too
Sure, DM'd
🙏 Thank you for your assistance, but know there's a common audio artefact in some v4 songs -- this likely isn't a mistake or misundertanding on the user's part
Yes, that's the basic extension, but I read that with V4, we could directly move to a longer song without using the "extend" option...
@limber pike
Does anyone know how Suno.V4 has increased wav quality.
It's less compressed
And Stereo space is better
Is the subscription real?
yea, seems V4 still has a lot of bad audio issues
The sounds themselves sound better, but no changes have been made to the actual .wav files
Right, It's even like tis on official suno examples for 4.0
Is the subscription real? Because I don’t know if it is
Well you cant use it commercially without subscription
It sounds like a metronome type percussion or some other instrument fx. Try listening to it on headphone or speakers and not your computer.
is v4 working for generating instrumentals? I created a song with V4 toggled, and the end result songs had the V3.5 tag
I might need more context...
<@&1150459415359664138> can i create remixes on suno?
Bro, I am a professional SOund Engineer listening on DT 770 Pro
hmm.... I'm interested in just how much I can massage things to get this one minor change in ("In defense" to "On behalf") on a Remaster
It’s in $ and I don’t use $
like i put a song and it makes me a remix
Welcome to "The Shimmer" 👀 -- you don't need to convince anyone that it exists.
It's in every 4.0 creation
Not just mine
The closest thing would be the “cover” feature but it’s pretty inconsistent with how well it works
you shouldn't tag the devs @tropic estuary did you not read the rules when you came in?
It's definitely a bug. The repeats of the sound are even at the same tempo across different songs
oops
Same pitch, everything
mb
It's not the shimmer
Was the Bitarate 1536
I just want to know if it’s not a scam
It's a legitimate company.
Maybe a stupid question, but anything could help at this point. Can you tell me what do you press to get the list of models to choose from?
I listened on Dt 1990 Pro.
That doesn’t answer my question
It absolutely does.
Then why are there so many bad reviews on there
They're referring to with a screen-reader
Not sure where you're reading these reviews from -- either way, caveat emptor
What?
Does anyone know if the Bitrate of the wav file has increased in V.4
it really depends on what your skill is with writing or what you want to do with the Suno tool...
use it like you're having a conversation or taking notes describing a song and let Suno write the lyrics and layout for you
OR
plan and go through the writing process then add the layout in to the tool and generate..
Play around and go from there, you can sort of learn as you go, or learn poetry formats and syllable counts and all that stuff... up to you how you use the tool provided
"caveat emptor" == buyer beware -- if you don't trust Suno, don't buy a subscription.
It looks like you've been using Suno for the last couple months -- presumably you've gotten to know the platform. If you have any direct questions, we're happy to give our opinion, but we can't be expected to answer "why are people leaving bad reviews" without knowing where these reviews are being left.
In there a way to merge two separate tracks/songs into one?
Not directly through Suno 🙏 Unless you mean stitching the extensions together with the original clip -- that's "get whole song" on the part 2 clip in ... > Create > Get Whole Song
Seems like noone knows anything as of yet
there's ways... maybe not yet in the Suno tool
Are you on a paid membership?
No Is free
V4 only pro and premium i guess?
Bolded as such at #📢┃announcements
Has the quality of Stem seperation improved.
can still hear vocal track in instrumental clearly
Like with the past models, newest models are limited to subscribers for at least a week before opening to free (if it will be). During this phase, the up- and down-votes from the subs are used for the final training of the model before it is fully released and frozen
NO. I have a subscription
It's a noise artifact it was present in 3.5 but was more shimmery muddy and accumulated thru gens. This is like that turned up 11. Wonder if it's gonna be looked at.
all things that launch come to subscriptions first. Just look at what people are complaning about with v4 right now. there's audio issues all over right now
refresh your browser, clear cache maybe
Try refreshing your page. If that did not work, try cache-bypass reloading (#📚┃suno-school message) or simply relog
problem: sometimes switching between create and library causes the page to refresh and the song I'm listening to to stop playing. it didn't used to do that.
me. I wonder why not put it for free
I used suno for like 1 day then went to the 2k credits plan... i wouldn't know what it feels like to be a free member lol - that's why it was a guess
Is there no way to use 3.5 Replace Section anymore?
lol I did it for 3 myself
What is 2k credits plan
I Need create big room Song
👆 yea that one
It seems to happen particularly badly on songs that begin with drums in the intro before other instruments come in. Like it can't handle the silence to heavy staccato right away and just delays the crap out of it.
For any screenreader users, I've found that it's possible to select the new model in Mozilla Firefox, but not in Google Chrome, so if you're struggling, try another browser. No idea what it's related to, but it worked for me!
I use chrome and had no issue, could be numerous user-end reasons, I would suggest some basic troubleshooting like, restart, clear cache, log out-log in
I get an error when using Persona and trying to generate a song -
v3 or v4?
v4
Suno v4 Do you think they'll put it on for free?
Usually means something in your style prompt or lyrics is being blocked by the automoderator https://help.suno.com/en/articles/3198209
It might be similarly related to the song that the persona is using as a template
We support a welcoming community for anyone to make music, but inappropriate content may get flagged before you can generate a song. At the moment, so...
yea usually doesn't take too long until it goes to free users
Have you made a v4 song outside personas yet?
Okay
Not surprised lol. Now I am generating with V4, but it generates whatever it wants, not following any prompts 😄
i have a question if i end my song with v3 does it affect the quality of the rest of the song (was v4)
ah, that seems to be the case, it didn't like the lyrics created by Suno itself haha
nm the answer is yes it affects it. it lowers the bitrate it sounds way worse
I haven't tried any fresh gens today, only remasters that... had their ups and downs, shall we say
I have a song written that I haven't generated yet I should try it out
yes its not as we expected 😩
so v4 is gonna be horrible at ending songs i guess heh
ill just find a place to hard crop
it happens, sometimes the AI simply doesn't get what you're trying to say/do
I had a song that I had to do everything manually because the AI simply didn't want to work with it, it generated ok in the end but didn't want to do any structure stuff... ended up being a great opportunity for learning tbh
remaster is going strange because it takes the idea of 3.5 and makes it to V4, if in 3.5 there are strange things the new version will also do it, but I suppose that if you do it normal it does it well
i cant create cover songsssss whyyy
Well - Remaster is creating really short clips 🤨 - anyone else got this issue? 9/10 are those 2-5 second clips and some are slightly longer, but I don't get a full remaster so far...
I wanted to try version 4
seeing as this is being a relatively common error, it might be a version translation issue, best to report the instance to the bug techs
Create my track remix crash Bandicoot
guys I need help!!! so when I try to extend a song it FORCES me to use V4!!!! I've been trying to click V3.5 or V3, but it doesn't let me. is there any way around this??? (and it only does this when I try to extend a uploaded audio)
that's interesting... I've remastered nearly 8-minute songs with an issue. try refreshing your browser?
The v4 badge is not appearing next to my songs even though I have it set to v4 while generating, anyone knows whats going on?
that bug artifact is super apparent in metal lol
It's interesting that I'm not getting it with grunge though
its in a lot of the metalcore
rock or anything rock adjacent it seems, i've got some soul-rock tracks that get the clicking quite predominantly
already logged in 2 times again - issue stays the same... Also: Suno does no apply the V4-Icon to my songs, even if I create a new one under V4
it sounds like its trying to do reverb, but completely wrong lmao
Seems like V4 not working for me rn
ReMi model seems not to be working, as everything that I'm trying to create for a test with ReMi, it gets stuck and doesn't finish. Classical mode works tho
should I try logging out and logging in again? or is it just something I can't fix? 😭
its a mobile bug, it does the same for me but with v3.5
iOS app or browser?
browser
i have an android so im not sure about the app
tried force closing the app/browser?
idk im using it on pc right now
I so badly want to use this to create stems to practice drums against but there's a huge issue I can't seem to resolve. It needs to be quantized at a specific tempo in order to make that work. I was told in gerneral chat that it's not possible, but I thought I'd ask in here as well, since it seems to be a technical issue.
dont practice drums to AI beats lol
you cant lock the tempo
so i wouldnt advise it
lot of promise though, and it would be a very cool use case; I'll keep an eye on it.
Sorry thought you said mobile
I'm trying to use "Personas" in version 3.5, but it automatically switches to version 4. Is this a bug?
Hey, I use "replace section" with v 3.5 a lot but now it forces me to use v4 and it sounds much worse, is there anything I can do?
it is
i was just answering someone else
not asking for help
god the artifact is such a strange thing to happen
makes me wonder how it develops something like that
It is compressing the shimmer but also truncating it, creating a loud delay effect
thats what I was saying, sounds like reverb majorly failed lmao
in the v4 Feedback channel there's a thread that's started about it about a minute or so ago
V4's singing in Taiwanese is weaker than V3.5's. Why did it regress?
Different model = parameters changes & AI learns different patterns
I've been waiting for v4 since 3.5 was released and apparently, even with v4, extending songs still absolutely refuse to use the prompted lyrics. Tbh, it seems even more broken now since in v3.5, u cld just change where u extend from and it hopefully fixed it some of the time. In v4, nothing I do makes the A,I use the prompted lyrics lels... Any workarounds?
it's not just in taiwanese, it's got issues across the board, but it does have some improvements in some ways also, it is still a WIP after all
Even though I select the v4 model for my song, the small icon next to the generated song still shows v3.5. Is this a known issue? Does anyone else see the v4 icon correctly when using the v4 model? Any help would be greatly appreciated!
I would recommend to keep trying and see if it's not just a generation in particular or a sentence structure that needs adjusting
most likely, i think i just saw something like that in there
ive spent like 300 credits trying. also created new instrumentals to then extend off further varying from 3-10 seconds. v4 just appears to b broke with extensions.
https://suno.com/song/d63fa403-deb3-4547-847e-e1e40ffb36b3 Does the 4.0 also have these sound artifacts for you?
Dark psychedelic classical hard rock and heavy metal, acoustic guitar background, epic eletric guitar. song. Listen and make your own with Suno.
It is overkill for storage, but not if the end goal is for distribution. I just downloaded a WAV for one of my tracks and it shows 16-bit/48 kHz, which is considered broadcast or studio quality and is slightly higher than 16-bit/44 kHz (CD quality). It is 03:17 long and would be 3 times larger at 24-bit/96 kHz at ~108.3 MB.
It would be expensive for Suno to offer the higher quality 24-bit/96 kHz WAV as it would take more compute time to generate from the audio array and would likely increase egress costs compared to the 16-bit/48 kHz WAV file. Perhaps, they could offer a tier or add-on package for higher quality WAV generations which would offset the cost of generation and egress.
can you timestamp please?
that sounds normal to me.. edit: oh, I hear the ticking..
it sounds like there's a slight double up like there's 2 drum kits doing kicks but other than that it sounds like it's meant to be there
no there's a whirring int he background, it's not doubled
i'll listen again
start at 1:12
https://suno.com/song/12301b9e-e3db-46ad-9be5-2073b2c7667b I have it here at 1:20 as well, right after the chorus
Grunge Female vocals song. Listen and make your own with Suno.
But when I used a Taiwanese song made with v3.5 for v4 to edit, it actually turned a song that v3.5 sang correctly into something incomprehensible.
ah yep, haven't had that one myself, but it's definately there, albeit it's a minor issue compared to most right now
idk, a music generation creating music artifacts is an issue
Could this issue be only in the remastering?
in principle, remaster is an ingenious feature. But I have the feeling that the tempo can be a bit off from the original and it changes the style slightly once in a while
Let me fresh generate it, and just use the persona
Is the website down? All generations get declined
it's fine here
for me it generats but i cant play them anymore
yes i've had the tempo go inconsistent from the original and can slow considerably which then causes other generation issues
I just got an error „Song stuck while generating” for V4, when remastering an existing V3.5 song.
thanks.. something similar is there. I don't know if they are still rolling the update to users
trying out how audio upload works with v4. Created a beat, I have lyrics ready, now I wanna make V4 sing them and continue the song after 2nd minute mark
yep, definitely slower, not faster
It sounds like reverb to me on my headphones. (ATH-M50x)
Note: that started happening after I was able to remaster another song.
Correct error message:
A clip got stuck while generating.
No credits were used and you can retry.
I hear a whiling tick-like sound when I listen on speakers
well, now it's producing text in the icons for the songs
I got the same problem i had last time with suno getting stuck generating clips, i was told it could be the profanity etc. So... now i got it again with a different song, but suno actually already generated multiple clips for this song, there is no profanity in it, why doesnt it want to work now?
It has nothing to do with profanity
phew! good to know. You arent the only one having the same issue
Check here for the issues with V4 many other users are having the same or similar issues
https://discord.com/channels/1069381916492562582/1308499132356362383
Question on Stems. Since the music was generated in Suno why do I not get a clean stem? I can hear vocal echo in the instrumental and instrumentation in the vocals. Shouldn't that all be handled on creation to make sure these are truly separate tracks?
welp
it should be back up. try again
Anyone having issues with Remaster?
numerous people, numerous issues
Aight so here is what seems to be happening.
always gotta rush features for investors - screw the users
cusom mode lyics dose not work give me diffent lyics
Personas weren't ready for primetime either
Checking.
Confirmed, I am able to remaster now
I have been remastering since the announcement happened
good for you
Thanks for the atta boy
im paying $60 I want features that work
Ok then read. Beta
https://suno.com/song/d9c3d267-e6cc-4ed9-a8d7-3a587e47e189 it's here too, even before the intro finishes
Grunge Female vocals song. Listen and make your own with Suno.
I noticed that although the sound quality has improved in v4.0, the song lyric recognition seems to have become weaker.
You aren't paying for beta
yes
So instead of complaining how about we do some learning
no but Im freely giving my time for them to test it
do you mean from written to vocal?
And then come in here with such constructive criticism I see. I am sure they are happy
again, paying for their service, premium, are you?
ok well you are more satisfied with a product making you a guinnea pig than me
I am just saying that complaining about a beta is missing the point
no its not, I work in product, I know how this all works
and they shouldnt release features that dont work
@foggy shale @worn flicker as much as this is a thrilling debate on both sides, can you take it out of the tech channel?
I got V4 working! It's cool. The constant audio glitch is definitely a bug because they said it's a beta for subscribers only. By the time it's available to free users, i'm pretty sure it will be gone.
I am wrapping it up now sorry
how do i put in a support ticket? v4 comes in as v3.5 and everytime extending upload maxes at 2min
Must be my old ears, but remasters don't seem to experience glitching with V4, except that fleeting issue with generation errors
@lunar anvil https://suno.com/song/0f0d79df-213d-43bf-a094-e560e324e252 brand new generation, it's there within the first verse
Grunge female vocals song. Listen and make your own with Suno.
So what I originally came in here for was things to note. I saw people talking about the "reverb" ish effect.
From what I can tell it is doing noise reduction on the flanger sound that would happen with the longer generations in v3.5
Well Hiss reduction
From my observations in here there's been 3 distinct issues with artifacts brought up
A clicking noise
A soft static/whirring
and the reverb issue
De-hiss
its very unfortunate this metallic echo delay sound that is in almost everything im creating. if it tries to do atmospheric stuff, seems to be the trigger. it's always the same frequency too.
100% always the same frequency
Haven't heard those yet. I jumped strait to my songs that had the most drift, and that is what I've seen so far. It cleans them, but it still has the original drift
yea it's a shame i liked the more atmospheric tone of my older metal and heavy rock ballads
im super bummed out, because it's finally getting dubstep right lol. but they all have this annoying issue! everything else about the track is beautiful. damn
you want an example? I can post both the original and remaster of one of mine, you can head it disctinctly
Are these all on remasters or new gens?
both
Guys, don't complain about a beta. I mean have you ever used apple's beta software?
add to that instruments morphing into other instruments. Timbres do not remain consistent
I complain about betas all the time, and betas get fixed that way
... this is the channel to talk about technical issues... we aren't complaining, we are reporting
The problem is is that there was a testing period. And they released it broken.
Just from being in here while testing was happening, I can say that it is known that remaster is not perfect
but that sounds fun and trippy
It's a FEATURE not a BUG
nah, this is still test period
It strait up said that, paid users get to Test.
It was a known issue before it was released to subscribers.
yea pre much
Nevertheless, still keeping up the optimism and waiting
Yes, you know how you stress test the beta? They were in Alpha before
that sounds more cynical than optimist
better not have been.
this is a huge issue tbh, a constantly appearing noise in many generations? makes every gen a waste of time
Okay it's like a video game has an alpha. Imagine if your gun makes an extra sound each bullet, and it's distinct and annoying and people report it.
But then the video game company releases the open beta on Xbox and the bug is still there.
Does that feel right or am I just too high for this chat?
(and my money)
its a beta because its unfinished i dont see how hard that is to understand
I understand that ffs
then for any generations that have issues use the report button and state the issue and someone or something will look in to it to fix it
I hear high hats far in the background that could be the ticking sound.
Alpha is for the non polished program just to test mechanics. Beta is when you let more people test a more polished product. Then you get to gold/pre-release, then release
Why is there a captcha appearing when I'm trying to create a song? I use Firefox and it won't allow me to do it.
Unbelievable. They release V4 but I can't create with it unless I use a different browser.
ctrl + F5 (to refesh the page)
Clears the CSS cache
that's not what I'm hearing.. but it could be Im using crappy speakers lol
there needs to be pins on this section....
Basic trouble shooting 101
and
a tag to the feedback channel
Yea, there are normally mods around, but no telling. I didn't know v4 was getting dropped today
I got a discord bleep and saw that it was dropped and did a few remasters and i've been in the tech channel for like 3 hours now....
Note: This bug has been reported: Just tried the new-fangled V4. SWEET! ٩(^ᴗ^)۶ But, problem: I added a word to the song and when I played it back the singers skipped the word. Is there a way to fix this? I'll try deleting my cache.
Oh I'm bailing until it calms down back to tech talk lol. I got sucked in immediately so I must not be in the right head space for this at the moment. I'll be back later.
how did you add the word? with replace section?
Anyone else having issues with songs being created in v3.5 instead of v4. V4 is an option and it’s selective. However, the song is created showing 3.5.
Thanks, but that doesn't work for me. I'm now unable to create.
FFS, I'm paying to use this. Get rid of the captcha!
I get the ios popup on android all the time, I feel ya
I am having the same issue!
have you had the captcha before today?
personally i've never had one
Yeah. It's just so frustrating.
I don't want to use another browser, but they're going to force me to do it.
1it may just take a few tries to get it that to go away, have a ittle patience with it
First time today. I can't complete it or get rid of it.
STEM seperation is never a clean process. If you are a paid user I would suggest download the WAV and then separate with an external tool in a DAW or a free tool (Splitter from BandLab).
I mean... is it Suno related then? maybe run a scan on your malware detector?
The way generation works produces a full track not separated. So, that means like karaoke, you won't truly get the vocals out of the instrumental as much as you will be able to get the vocals isolated. The frequency of vocals hit the same frequency of the music too often.
So far on remaster the dehissing done does make it sound a little like a lower pitched flanger wobble
I'm confused. They really need to make some documentation. People can't use v3.5 replace section and now (I could before) I can't use v4. The clue may be that the song is over 4 min long.
does it actually stop you from generating that though?
I could never modify a 4+ minute song.
The cymbals problem in many many tracks, some cymbals/ hi hat are really repeated and fast in bg.. Anyone has this problem ?
egyptian filmscore, drum and bass song. Listen and make your own with Suno.
I am glad we can remaster them. That tells me it acts more like cover.
if you took a quick glance, you'd see we're all complaing about it https://suno.com/song/c19b6028-c681-410e-a012-d92bc550d677
electric grunge gritty song. Listen and make your own with Suno.
yea it's basiaclly cover and persona in one for the whole song
I guess I can try to at least see if it actually works. I also thought we can only use it for songs <4min. Even if for >4min I can use v3.5 ... write it down somewhere. This is more and more complex and I really do not envy new users.
Yup. Overall not bad at all. I am going to have to learn how to prompt it. I got way too used to 3.5.
I can remaster long songs
Yup
was a statement, not a question
Honestly it sounds fine, but I mean i have tinnitus which doesn't help
Oh yep sorry !! Glad to see i'm not the only one
Oh they were responding to me not asking either...lol
I selected "Edit Song Details" in the lyrics I added the word twice, once to each of the 2 chourses. I can't figure out how to add a word to the lyrics using the replace. Pretty difficult.
it's irritating because vocally they're much better, the instrumentation is BAD..
The instrumentation does sound like it is overly tight, just off the bat as I hear it. Like there is a lo-fi that is missing to fill the gaps
on mine?
Oh no, on all
ah..
that's adding the lyrics only to the video generation, not the song
Bad news is it's extremely hard to get the AI to drop or add a word and keep the original as it was
they're adding things that need not be added
Everything I have done so far. It is too quiet in between pulling out the natural reverb.
what's even wose is that prsonas don't care what changes are made, it keeps using the original ones. even in v4
same. iunno, the bad is outweighing the good for me on this release.
you sounds like all the people in the star citizen reddit and discord
Oki-doki. Thank you for clarifying. Honestly, that really helps. Thank you. Much appreciated. It's actually no biggie. Glad to know what the Suno can and can not do.
Again. thank you. ٩( ᐛ )و
you listened to the remaster version of crimson threads, right?
yeap totally annoying, I don't use Personas for this reason
welcome to the #📚┃suno-school channel? 
https://suno.com/song/6b7247d9-9530-4d9f-959c-3665cbb7a6ac here's the original for the record, I was not refering to the personas with the things that need not be added
electric grunge gritty song. Listen and make your own with Suno.
Do we need to create personas again from v4 versions of songs or is it automatically adjusted to v4?
they are fingerprints, so the personas are fine across models
which things are being extra here?
listen to tthe remaster after the original
lyrics, instruments?
instruments
stand by 7 mins
btw nice homage to threads of destiny from japanese lore
hm? where?
Ho ho, nvm then
LMAO
yea, That's what I tend to call a "demo song" It's so I can listen to the timing that suno will do with it
ok i see what you mean, the cymbals are much more pronounced
I have lyrcal work to do with it still
"The darkness lingers, in the dark." < this is 100% going
and btw:
For the Japanese, human relationships are guided by a red thread that the gods tie to the little fingers of those who find each other in life. Legend tells that the people connected by this thread will share in the enacting of an important story that cannot be thwarted by time, place or circumstance.
oh, you want to know the theme of this song? Crimson threads would make more sense when you hear the basis
I don't need to know the premise in your particular circumstance, it was just something I picked up
yea, the connection was accidental
hey guys. aren't covers supposed to keep the original melody of the song you are covering? at least that was the case for me until yesterday. But it seems they are broken now and they keep producing random melodies. Which makes no sense cause then I could just reuse my prompt instead of Covering...
It's actually a thread of songs about 2 specific characters
but yes the Cymbals are much louder even drowning out other instruments etc, I found that regen can dull it down... not an optimal solution but if you report on the failed ones it may help
Is there a way to use covers and personas using v3.5 instead of v4? I'm really not liking v4 at the moment.
https://discord.com/channels/1069381916492562582/1308499132356362383
Check in here, the mods aren't on atm and I haven't experimented on this issue yet
You need to first creat a song under 3.5 with a persona. Then you carry that song over to v4 through a cover. Thats what i have been doing.
But the thing is: I don't want to use v4. It sounds awful, and gets the voices all wrong for the style I tend to use.
I agree they sound a bit more generic than 3.5
I hope they revert it before it becomes an "I want my money back" situation.
Keep it in 3.5v and use a cover
It won't let me. I've tried and wasted credits on this.
it already has a couple times in the last hour in this channel alone
It very well may with me
In what way? i have 3.5 songs i started with a persona. after i find one i like and fix it up a bit, i carry it over to the cover. Its the only way you can do it.
not what I was talking about
woops
Guys, am I the only one who noticed that new generations and remasters of old tracks are subject to reverb and distortion?! It’s like this in every track, and an incomprehensible echo that appears periodically
So, would creating the voice you like in 3.5, make a persona, and then make a 4.0 with that persona work. Worth a try?
i wasnt talking to you. i appologize
The moment you add a persona it forces v4
4.0 and personas are not great at all
it's a solution, results may vary due to it's AI generated however...
you can change it back to 3.5
I did. It didn't work.
The new version 4 cannot use old persons in the form in which they were written in version 3.5. It's very sad because these are different voices.
Lately I've been having songs (v3.5) that just stop in the middle, as if cut off. Always in the range of 2:50 to 3:20 minutes. Has anyone else noticed this?
yep
😮 well then, i tried and it doesnt give me an option. so i guess so far you're stuck with just 3.5v till they allow it.
I've used musical to get duets and "isicathamya" [African style genre] in 3.5 and get voices singing over each other, but that's more call and response.
well i guess i'll give up on covers until they are fixed.... RIP Enthusiasm 😦
I just did a track and there is a woman in one verse and a man in the next verse
Yes, I get a long pause then some background hard-to-understand dude doing a vocal rappish off the lyrics then finally glidding back to the orginal singers. No biggie for me. Plop the song in Filmora and cut that part out. A-batta-bing-batta-boom one complete song with no breaks of gloomy takes.
Cool. The trick is get it consistently, I guess. Good job.
mixed vocals are rare but they have been possible since v3
Hmmm interesting!
I hear voices, but I think that's just me.
you could try encouraging it through certain prompting
You for sure!
Just signed up a couple of days ago and a couple of my songs in v3.5 have beautiful duet with a man and women. Totally awesome impressed.
when something you prompt comes together in an incredible and/or surprising way it feels so good
I'm just after that great beer song. This is my chance.
Think I might have to continue making my tacks in 3 and then master it through 4
getting some strange results with layered harmonies/choral parts translating in remaster to v4... feels like its overemphasizing vocal clarity and isolating certain parts more than others...
🎶 I've got friends in looow places 🎶
V4 is wild animal suno fix this cover are useles and also persona
I've also had issues where my back up vocals are too varied compared to the main vocals, only noticed this change in V4, V3.5 was fine
are you noticing some pitch variability too? It's hitting the melody it seems
Is there any way i could recover a song that got an error and disappeared?
Did I hear " .... varied backup vocals...." perfect crowd response for a beer song. I'm getting psyched.
It generated the whole way but kept loading
absolutely, pitch and sometimes tone is completely off, depends on the gen, noticed it on my remasters
They can just fine. I'm using V3.5 personas
my personas can't be neglected
for me no one new rework had with original 3.5 persona
all in all it seems that the concept of remasters was genius, implementation however... not so good, hoping it will improve in the future
I won't be using the persona or remaster features for the foreseeable future
Lyrics seem to be the biggest improvement in the new model... loving the ReMi
we need to wait more than one year to have a really good quality of a product, as I see))
possible that it's just a V4 issue, I haven't generated an original on V4 yet... been too glued to this channel lol
My intial thoughts on V4 prerelease:
The 'shimmer' as the community is calling it, is too much (at least in my usage). It is way too consistent, and if you're any bit of a music producer / audiophile, its very jarring.
I am hoping the fact that it generates at seemingly the same frequency in every instance of it happening, that it will be easy for the team to isolate and fix.
It is currently striking me in about 80-90% of generations. This is specific to my experience, likely due to the styles I am prompting.
As it stands, this makes a lot of my gens a complete waste of credits, persuading me to not use Suno right now.
Which leads me to the next point:
Another issue is removing access to older tools. Persona and Cover now strictly using V4, isn't great. I don't see how it would hurt to keep it functional and still have version be selectable between 3.5 and 4? I would have kept creating, had I been able to go back to my original workflow, but I can't do that either now, at all. This hasn't really been a thing in the past. I mean we can still use v2, even. More tools in a creators kit, is a good thing. Even if they are old tools.
I like the vocal quality but not the voices lol
Ok, next observation. I get why they tightened everything up to stay away from drift as much as possible.
Yeah i got 5k credits i wanna wait
if I use 3.5 personas in version 4, this new tracks are very different, the voices are different
You know how Vocals are night and day difference when live and they have Reverb on vs off?
same problem
It's just a whole different person for me lol
Oh dang
I'm just gonna master my old instrumentals instead of play around with the vocals at this point
truly an honest take, thanks for sharing
I perhaps have a different... possibly more wasteful approach than you do so it isn't as disruptive that I can't cover or use persona (which i never liked)
If it turns out that V4 is indeed the more boring and safe but makes more higher quality and consistent songs then i'm happy to go back to the wild west of V3.5
reverb distortion and echo, in all iterations, remasters, new tracks, etc.
Ok so it seems like they made these changes to stop or avoid the drift as much as possible. Which means the [Reverb] and [Reverb Heavy] tags are now back able to be used without a tragedy of mess at the end of the song
I believe it is a prompting change we will have to pick up on
I had stopped using Reverb and echo all together
I don't like reverb fuction
I didn't in 3.5, but in 4 that is what is missing. Why Vocals are falling flat
On the contrary, v4 seems to be working really well in the niche spaces, as I've been noticing with my own creations. I do a lot of weird gens with Suno naturally. So I am very happy with it finally really nailing down the obscure soundscapes I was hoping it would create. However, that being said, they all have this jarring sound artifact (the shimmering), rendering each generation useless.
well they were aiming for consistency... I guess they nailed it... in some way
It is also true though that v4 is more tame or 'boring'. Especially with the mainstream sort of genres like Pop or Country.
The Shimmering sounds like a new horror film
we need some button to turn off this Shimmer)
This shimmer is it the now lower pitched wobble?
hey, i'm premier and the song i create using v4 are still tagged as v3.5, is it a display issue or something is not working for me ?
Like left over cymbal hits
i would argue that it simply should just... not exist?
Agreed.
I've found with v4, the soundscape is almost always much clearer than the former versions, however, it loses it's atmospheric depth. It feels flatter in a sense. While also feeling wider somehow. It lacks the little details.
shimmer i think is the best onomatopoeia for it, because it sounds like a high pitch shimmer
exactly
when i first heard that shimmering sound i kept thinking there was a phone ringing at work lmao
100% this, it was my first impression of the 10 remasters i listened to
I'm most likely not going to use V4 at all
Right on!
its like a consistent echoing metallic delay. you can hear it throughout this track, the biggest example of it being at 1:37
https://suno.com/song/12d2665c-2e47-496f-85b8-56cfa7d0dbba
Not to be confused with the hats, it is not hats. This same echoing sound is showing its face across different tracks. The same note and frequency too.
Seems to hit more often with atmospheric stuff.
low frequency, deep bass, dub, wub, wobble, future bass, minimal, industrial, deep dubstep, drone, deconstructed, ambient, taiko, melancholic, sparse, foley, 140 bpm song. Listen and make your own with Suno.
Remaster is going to work differently but hey all versions are available so we get choices yay!
it look like this sound it dead withnut some details
ewwww I hadn't heard that yet.
is there a way to cover a song that you did not make? I'm trying to cover a song made by a friend but i have pro and he does not and i wanna use v4
it sounds like its ready to feedback
Yup, that is what I am talking about. That sounds like de-hissed drift.
And for some context to readers, I want you all to know. I've been a music producer for almost 13 years. I am using quality listening equipment. My ears are trained... and I am an experienced Suno user. I'm not coming to you with these takes as a noob here. ❤️
Hey @humble pawn adding some stuff to server, can you check and see if you can see this channel? and will it let you add a message?
https://discord.com/channels/1069381916492562582/1294418290348921007
What is Remi? Never heard of it.
anyone else having trouble changing to 3.5?
my instrumental artifacts sound more like someone left the drum clicks on and they make it into the stem
I try to change it won't work
can see, can't type
Thanks
can also react to the post.
Basically the same rules would apply. Create in sections. Make Tempo changes to get AI to check its timing.
Also thank you to everyone in here helping other members! It's incredibly thoughtful and helpful ❤️
So with that same song, see if the same thing happens if you do an extend from when it starts to really pick up. The sound obviously
The same commentary was made in the shift from v3 > v3.5; my take on that is that as the technology evolves, it’s going to evolve towards producing the average / typical sound, which means that it’s going to be easier for someone doing basics to get something basic, but you’ll have to actually craft and create with prompts moreso to get a more unique sound. Which, nothing wrong with both approaches! But does require digging in more to get more out (which imo should be a good thing)
I want to see if it is actually drift or a new artifact
I listened on the DROP BMR1 speakers and could not hear any ticking either. I have noticed that V4 does have more nuances. It could be that our different acoustic environments are influencing the sound. ¯_(ツ)_/¯
flanger wobble?
From what I can tell, it is almost too clean...if that makes sense, it is missing the natural Reverb that happens and any Reverb that would go on due to acoustics. So going forward I'll start using Reverb and echo tags again
v4 best vocal only prompting?
Anyone have luck with that?
maybe I should grab my phone earbuds, that would tell a bit more lol
Is it technically possible for Suno in the future to have the ability to make full songs without drums? Or be able to separate them in stems?
From a drummer's perspective it would be fun to be able to play to songs and add my own drums to them.
You can separate the vocals quite in Suno. You can make without drums specifically.
but does the generation listen to your prompt 90% of the time no....
Any commands in particular to get that to work?
Im gonna try A cappella later and see how clean it is. Lets hope the simmering sound doesn't happen
Yeah, that’s one thing I noticed with my remastering attempts - a lot of the songs I made had some amount of reverb, but if reverb isn’t a style prompt, it seems to just immediately get rid of most/all of it in the remaster and so everything sounds more “present” but also less “produced” (in not-always-the-best way)
I don't know yet, I am talking about all versions
No a tag
I’m gonna try manually re-tagging all my lyrics when I get home from work and see if it majorly effects the remastering outcome
barbershop will get you harmony vocals
I struggle to get it to prompt specific intruments let alone isolate or not have one in particular... but I guess I haven't tried it 1000 times and maybe it was gonna be on the 1001 time (i'm not being sarcastic)
I haven't tried yet to be clear, I got very used no to not using Reverb tags all together since 3.5 got messy. I am just saying that will be what I try out going forward. Like in v3
Yeah, when it comes to rock music, for example, I've never gotten it to do without drums. Excluding styles doesn't work. But doing brackets [ ] during the song for things like Electronic breaks, breakdowns etc. have had great results... Wonder why it can't separate instruments but can do that stuff...
V3 did that better than any other from what I have seen. I don't know about 4 yet
yea the breaks I was able to prompt reliably but not solos like in V3
I hear ya, too much time in the discord and not enough with the tool lol
There are definitely differences. From what I can tell I got so used to the wrong sounds that came from 3.5 because my style selections are way off in v4, more like it is closer to what the Explore page shows again.
Some instruments needs another keyword for higher reproducibility. E.g.
- For sitar, you want india or Indian present, otherwise it will sound like guitar
- Jingle bells loves Christmas
- Theremin + Halloween
- Violin and... Orchestra I think? Surprisingly hard to get it with Violin alone in the past
Note I did not try any of these combos in v4, it was in 3.5
Hey was that us doing sitar stuff in 3.5...lol?
I am getting SO much fuzz with v4, almost worse than 3.5 😦
Yep lol
Why can't I cover a v3.5 song straight into v4 anymore? I could earlier today
guys, add tone scale in suno please, i need this
Yes agreed, also the vocals seem more expressive and dynamic, however the quality is still very unlike the clip on Twitter for most generations as of now
Unfortunately, the shimmering effecg was a known persistent issue, more significant in the loud genres (particularly Metal, EDM/Dance, and Rock) as you can see over at #1308499132356362383
The devs tried to stamp it out but, well... guess they decided better to release as is rather than delay for like a month
thanks for those tips, I had sort of figured out that the styles tied to the validity of the prompts, it just took me a long time...
I am currently getting fuzz creating lo-fi ! 😄 It's not very loud at all 😄
Hopefully one day it will be a blast from the past
v4 has a problem with sound panning. vocals are in front and instruments are somewhere in the ass
Ah true, rephrased to emphasize it is more noticeable in those genres I listed
I do hear that
I try to keep them out my ass but they go that way...lol. I use my headphones to avoid sound assage
@daring lava the updated docs/guides/resources category is phenomenally compiled 😄 💪
Thank you! MikeB did all the real legwork at help.suno 😅 I just pointed to it
So, I had a song I was making just before v4 dropped, and this was an immediate tell on the style differences between 3.5 and 4.
Since it is speed rap, it had slowed down immensely when reusing prompts in v4, so that tells me that v4 must be back closer to what is on the explore page.
everytime I try to create with v4 enabled it produces a v3.5 track - same with remastering even though I click on the remaster with V4 button
there is a problem in the remaster. If the sound is initially bad, then the remaster will not be very good. The vocals will be better, but half the instruments will simply disappear, and these instruments will end up in the ass
same problem
I'm hearing what you are saying as we speak
Does anyone here know the difference between the two generations? they seem to be consistently different with vocal and instrumental pan
Has it been stated anywhere how it's quantified? the first generation feels like pro vocal and the second feels pro-original version/instrumentals
Has anyone else had this experience consistantly with V4 remasters?
Yes. I notice this
However, I have been noticing that of the two generations that one would be the maybe and the other was trash. In 3.5 and v4. Almost like one is to throw out trash
I wish I could relate, all mine have sounds artifacts that were never in the originals or the pan diff is abhorrent
I just deleted the ones I posted of v4 remaster. They were awful. Testers did say that remaster did have its issues you can see it in posts above. Maybe Jonathan was talking about it.
So, is all of the beta features going to STAY pro and premier? If so, that... sucks.
hmm, I guess it's hold on to your V3.5 and see if you can generate something you can enjoy on V4
I am getting loads of blank remasters
welcome to🤝 capitalism buddy
might want to see a doctor about that... sorry i'll show myself out now
Not sure if anyone else has this, but in all V4 generations I had there is a background sound, the best I can describe it is like a faint fax machine sound, especially when the song picks up
Way too far up there lol
literally everyone, you're not alone
yeah, i hear that and interference from a telephone call
ah ok
V4 is awesome.
I'd recommend people look at https://discord.com/channels/1069381916492562582/1308499132356362383
Unsure if the devs or mods are replying in there but basically every issue that has come up in here today has been raised in there
it has it's challenges, but the vocal clarity is amazing
I do wish there were free remasters...
free things are always nice
oh f*ck, hi hat sounds awful after remastering
V4 has the same problem as V3.5 of repeating versus. 😦
depends on the generation... if you generate a few times it will change eachtime
not a great solution as it's quite wasteful
the site is crashing all over the place on Chrome, windows 11 today, "Application error: a client-side exception has occurred (see the browser console for more information)."
Yeah the vocal is pretty good. I just tested a few songs. The instruments of the gospel songs are pretty good.
every generation costs 10 credits,, this feature works badly than cover for now
Wel, here's an example of a bad/poor remaster, including out of tune vocals. https://suno.com/song/4e57a25a-a0ee-47f4-8ffa-8e1e5b5ad12f
pop rock energetic animated song. Listen and make your own with Suno.
Hi @worn flicker, what's up.
Remaster probably should not be leading the charge on v4. We need to prompt it. Remaster is doing what it does but that isn't a good test of the model
that intro was warped lol
Waszaaa
I think the keyword is monotone
i've had similar experience with remasters
So... That's a yes?
dont know, hope not and it will be for free
Man I looked in v4 feedback. People are strait up mean considering they have no clue how to make these things. No wonder things take so long to get released look at the reception. Even when it is said open for testing. People talking about canceling because of a test allowed because they are paid users. Craziness.
yea I agree, we need more time to look at it from an original point of view and not judge V4 based on one new feature that will be fleshed out and improved on in time
If you spend as much time on the road as I do, you get to learn that people don't read even the most obvious signage...
I'm not surprised people are walking in thinking it's a finished product just because it has a premium service
Can someone explain this!?
Epic Rock, female vocals, harmonica song. Listen and make your own with Suno.
Very happy that I got V4 now...but I'm noticing odd "artifacts" on remasters. there's a weird "ohhh" noise before she sings a line. and some pops that weren't there before.
I think it happened back when they decided to start using beta as kickstarters on games. Making people pay for beta was a poor move. Early access is a much better phrase.
It's basically just a self admin crowd funding venture to help pay for development of their tool, which I wholeheartedly support
Do you think they don't have a testing department? Imagine you work in a department and you hear all this crap and tell your boss “v4 sounds great, we can release it”
yea Suno is basically a self run patreon
I'll gladly pay to test this and help improve it. We are still in the crawling stages with Suno. the future is looking good though.
Yeah I saw a post mentioned the voices are different. How many voices are their in suno anyway? They all sound pretty much the same voice to me, even V3.5. But V4 has new voice
someone is using api of suno ?
No API
If you want to get banned, sure. Suno has no official API. Services claiming so are actually breaking the law in USA by essentially hijacking Suno's Service
i have V4 selected but when i create new tracks it only creates v3.5 , i tried from the mobile app there it works V4. Also when i remaster it remains a v3.5 anyone any idea?
Yea, Suno staff just needs to bring the right things to the front. That is the main issue. Remaster shouldn't be what is being advertised. It should be create with v4. That is the way to avoid all the insanity that comes with recreating tracks that people have listened to 100 times. Any difference will be heard. Instead, no remaster until later and let people prompt and play.
I've counted 4 different female voices so far. 2 male (death metal only)
Anybody else notice the V4 songs slam all the way to the 4 minute mark without finishing the lyrics?
I remastered beyond 4 minutes, I was surprised.
Remaster sorta works. it sounds like its a remastering LOL. I can hear more and her voice for the songs i've chosen is close to what it was, just cleaner. So far I'm impressed.
That said, there are issues but again man, this is brand new. watch how this thing shapes up in 6 months.
You're gonna be able to remaster the original length so if you put a song together and its 6min let's say, its gonna remaster 6 min
at least, it has for me
Agreed. It is taking the original and tweaking, but like Star Wars, you will have people so used to the original that nothing will match up.
Cool. I am testing out Gospel genres. I wonder if I can prompt it to sing with black female vocal. Doesn't seem to work 🙂
Every new song that is not a remaster is 4 minutes but it never finishes the song. The prompt for this song is 2647 characters. I will play with it and see...
I doubt race will work instead of style
Hopefull it get better soon. Still new.
are you using an outro verse?
Ah. I guess, instead of race, put the style that the race sing maybe.
Yes.
some people also just put END
Jazz, R&B, stuff that would be trained...lol. I doubt they put black in there.
in square brackets of course
I personally use [Outro] and it works maybe 70% of the time
That is there too. I am using a previous song and seeing what V4 can do with the same prompt from V3.5. Remaster gave poor results so I thought I would see how V4 would interpret the lyrical prompt.
I've never put END lol
[Outro]
...
[Fade out to Silence][End]
You don't have to
Sometime [outro] alone doesn't work. I use
[outro]
[fade out]
[end]
It is all about inference
If it doesn't want to stop then start throwing down tags...lol
Same with vocal gender
I think also depend on the genres too?
yea I use male or female vocal in tags
Personas are weird. They are all shorter generations on V4 than on V3.5.
or punching the monitor, throw your phone at the wall 🙂
Definitely, it will always be trained on these things so whatever is the most popular will come to the front
That is a good way to get rid of a problem and make a new one...like stop smoking and start dipping
Here's a strange bunch of tags that ended up with something interesting
(i've been pretty prudish with tags until recently)
Atmospheric Rock, dark, War Drums, male vocals song. Listen and make your own with Suno.
Does personas suppose to sound different or the same. Most of my personas sound exactly the same as the original song.
Each model will be different just like in Stable Diffusion. The training is different
I've found this also, it's why I don't like to use them anymore
That I can't figure out either. when you get lucky and find something you like. make a persona
Persona has made awesome stuff and a lot of not awesome stuff
Yeah same here.
I would say supposed to sound the same(ish). I have one generation that has the persona changing gender or maybe it is more like a unison harmonic type. I can hear the persona being overlapped by the different voice. Very weird.
Yeah. I stop using them too.
With persona it has a copy of the original song under the new one, influencing the new, so if you have lyrics that are too close to the original then it will poke through.
Sometimes it will just poke through all together
i've had it bleed thru on a number of songs
When you change the style from the persona that tends to make crazy stuff happen
[Queue bleed it out]
Do someone have the problem that suno using v3.5 even when I choose v4 ? It's was working before and now even when I click on remaster it's using v3.5
Lots of screw tapes could be made from style shifted persona
I am not sure, but since v4 is the only remaster then it has to be using v4 to do it.
It's always show me v3.5
If you started a song with 3.5 and tried to extend with v4 then it would pop back
Even on my v4 generated music
Yea, what I mean is that there is no 3.5 remaster
At some reason I get a constant error
Can we not remaster V2 songs? The remaster option doesn't appear in the create menu for V2 songs.
then there's the answer... sadly
I have the opposite issue, I want to use v3.5, but it forces v4
Remaster is only to V4 quality
howdy yall anyone else have a heck of a time getting your personnas to sing? i keep getting an error message that they dropped the track.
I had this problem last week where I wanted v3, but it forced v3.5. Ended up being a lyric prompt issue where the prompt was over the v3 limit.
I mean when creating a new song, I select v3.5 model, but the song comes out with v4 badge
darn son... you writing a novella?
Sometimes my lyrics reach near the limit. I usually have to split or the AI sings super fast. At least it did on v3.5... v4 seems to have the opposite problem.
try makign a v4 song before using a persona
and are they normally 6+ mins each or are they quickly vocalised?
it fails on v4 too
I am getting loads of errors and repeated requests for captchas
It's look like the model we choose on the dropdown don't work, it's use some mysterious value
On v3.5 they were quickly vocalized. So, I would split-generate the verses to slow down the vocals and the overall tempo.
Someone else did mention the captcha thing earlier, I wasn't convinced that it was possibly Suno related as I've never had to do one for Suno
But if you are also it might be a thing
Looks like it happens only when using a persona
maybe a deleted/trashed persona?
good luck getting quick vocals outta V4 lol
Guys, just a quick question, did they changed how edit/replace works? before using it wasn't taking any credit, now it takes 15 each try
Thanks. 😉
I tried another one, and it still uses v4
I'm not using persona
Restrat PC or mobile device, clear cache, close app/browser properly, log-out/log-in and try again
The free replace wasn't suppose to last as long as it did, should have ended mid-October (see the bottom of #📢┃announcements message)
The amount being taken doesn't sound right and may be due to a visual bug (one that appears now and then). Reloading the webpage should show your correct total balance
Oops, forgot I replied to you in another tab
Anyone else noticing with v4 that with any rock-type songs there seems to be this warbling low metal sound? Sounds like a tremolo clicking
i'll continue here, guess it's more correct as section
Suno uses Cloudflare and it will show a CAPTCHA to users when it detects suspicious user behavior for a user's IP address or rapid requests which will trigger a CAPTCHA. Are you using a proxy or VPN, connecting via a public network (ie shared Wi-FI at a school, business, or school) or browser extensions?
closed the browser after logging out, relogged, 9760 credits, recreate section 1 try , yeah now seems ok, 10 credits ( @hollow hare , 10 credits is right?)
@knotty skiff just created a v3.5 song and it finished the song under 4 minutes at 3:42. Selecting the version was the only change. I wonder if the style of music is what is tripping up v4 on the tempo being so slow. 🤔
anyone else have an issue where generated clips wont play until they are fully generated?
yes me too
I would suggest not to play until they are generated. Sometimes you will get artifacts or the song will just vanish in error if playing during generation process.
always been able to before v4. could be an issue with heavy server usage. ❤️
Now I have to do v4 generations just to see if this new sound everyone is talking about is still just the drift with a low-pass filter over it
was curious if it was just a me thing
I am getting the same suddenly, no VPN - I am getting site crashes all over the place too -
I have had this problem since May. After losing a few songs I just wait now.
Hmm am i wrong or i've read 1 free remaster available on the new banner on the right?
I've just subscribed and tried 4 remaster and i got 20 credits removed 😅
if the site is malfunctioning, just wait, come back in a few minutes after a refresh. v4 has been awaited for over half a year, the server is being bombarded with usage
That happened to me once. I noticed that before it happened the sight kept trying to load over and over. It eventually gave me that check.
the free remaster is for free users.
ok
Best advice here, just wait until tomorrow
well thats just madness 
my suno addiction D:
Why is there an audio watermark in every song that adjusts w/ the pitch?
The laser beam sound?
Yea.
i hate it
That's def a watermark. It's also a subtle brainwashing technique the CIA does
thanks for explaining I was trying to generate like any other days and throws error no vpn set
Yeah totally. It's all i can hear atm tbh. Overall 4.0 seems very nice soundwise but that laser beam trademark sound is driving me crazy.
i grew up with those 90s laser toys from the faire the only reason that sticks out so much tbh
Its also the same sounds italian regional trains make. Trust me.
If you are on your home Internet try restarting your router. See if it pulls a different IP from your ISP. Also, you could check your public IP on MXToolbox to see if it is blacklisted anywhere. If using mobile device switch to Airplane mode wait a few seconds then back.
Do you have an example of the laser beam sound?
What's the style?
rawstyle song. Listen and make your own with Suno.
very noticable there.
Neoclassical Trance, Glitch, Stutter, Modulation Variation, Hypnotic Loops, Tuned Vocals
once you hear it you will never unhear it, be prepared.
Without it being a remaster
would it work with or without more or less tags?
Any dev team awake at the moment? XD
Anyone use a persona they already had with V4? and what were the results.
like being more or less specific
from what the dev notes state it upgrades to V4 as a whole
These are the same tags I used in v3.5 without issue.
https://suno.com/song/89c724e2-e1b3-4788-b731-8829d2146fd6 this is in italian so maybe it's easier for you to focus just on the music, from around 30 seconds, the instrumental part.
cyberpunk, synth-pop, electro, electronic, synth, synthwave song. Listen and make your own with Suno.
https://suno.com/song/c1e24bf9-7940-4044-a0a1-9a0bebe9f91a
Here, just for you
If it is a remaster then the sound you hear is a filter being put on old high end drift.
so your theory of it being something about V4 would be correct in a way, but it might still work if you're more or less specific
Yep, I hear it. Same deal. It is something that will need to work its way out. I am not used to hearing it off the bat, but your song started loud off the bat so it makes sense
I have done more beep boop sounds than I want to admit -- and have 20 years of sound and visual art programming. That is a watermark -- it's too constant and doesn't fit.
Not a watermark. I know what you are talking about, I do not have the same years in professional level sound and visual, but I have just as much in the field in general.
This is different
if not a watermark than what?
It adjusts with BPM, and Pitch... is watermark
any genre -- same sound, any lyrics... same sound.
You can probably find the frequency if you wanted lol
yeah i confirm that, remastered like 40 songs and is always the same sound
I'm not going to argue with you, I can simply say You are trying to apply analog stuff to digital stuff. In audio AI generation there is a possibility to get left overs from the encoding process. The track is created all in one. Since you know about encoding then you know what I mean when I say quality vs speed.
Oh, you could be right but the consistency is what makes me not think that.
How do I use the crop song feature? I'm thinking I can't remaster this V2 song because it's 2 seconds over four minutes.
I can tell you that it isn't a watermark but I can't tell you what exactly it is.
I see the select edit boxes, but when I type the time I want to keep it doesn't let me. How do I type the values?
There are watermarks out there but it is obvious
Could be is enough imo. Copywrite tools are sensitive for a reason.
Well fact is i feel like the dude in A Clockwork Orange. It's giving me allucinations.
They make theirs obvious, it says use suno and things like that
It happened when sites were using suno and making money off unauthorized use. So they have watermarks and they are not subtle
It sounds like a typical synth effect found in electronic music generated from a recreation of a FM synthesis, a subtractive synthesis like a sawtooth or square wave that has a low-pass filter with high resonance and then pitch envelope applied to give those zaps.
The style of music tags often includes laser-like zaps so the AI may just be trying to add them based on its training samples.
I have it for every genre, even metal or jazz
mmm no... not actually. "General Commerical Use" implies no watermarks.
I have generated some by moving the tags around. I noticed at least in v3.5 that if you move the order the AI weights the tags based on position. Let me see if v4 has that same quirk.
I'm talking about Suno and their watermarks. They are not a subtle noise they are in your face.
Hi, I have the same phenomenon. It appears when the music increases in intensity. I have it more or less on several styles of music, it's sometimes subtle, sometimes more pronounced. But actually, I've heard other redundant artifacts on the V3s. Less remarkable but present practically everywhere. I don't think these are watermarks but rather "artifacts" specific to Suno.
I noticed that about V3.5 also
It's not about the noise itself, it's that it's there at all. I pay up for a year, all of a sudden I get those watermarks -- that are audible?
@manic tiger try something like the following to see if you still get the zaps.
ambient minimal analog warm mellow ethereal melodic chillwave downtempo retro-futuristic smooth textures dreamy layers
lush pads soft tones organic synth flowing soundscapes vintage-inspired progressive relaxing melodic harmonic spacey
i will try tomorrow thanks for the tips. I'm out for now. See ya.
The noise was also in 3.5, it just didn't have a lowpass filter. I can only explain so far, if you want to call it a watermark that is fine boss. Talk to support. I am only trying to save you time.
Try this GPT as it will help you identify different tags and you can even ask it to deconstruct songs to give you tags to mimic the sound. https://app.aiprm.com/gpts/g-utsMc0LBR/essence-explorer
It's just how consistent it is -- we'll see whether it goes away. You could be right, I could be right. is what it is. 🙂 still a good tool either way.
"watermark" implies it's added intentionally
I can tell you how to counteract it if we get away from the whole watermark thing...lol. watermark isn't something that we can do anything about
I can't tell you how to fix it completely, but to minimize it.
Oh, I figured it out already. Something something negatives.
but how do you know? that's like trusting telemetry 100%
There were inaudible watermarks in Suno works for a while
Like at least half a year ago, most likely longer, if not even at the beginning
Not quite sure what you're asking -- how do I know that the shimmering / audio artefacts aren't intentional?
Which is fair; and they could still be present with the current model. Precedence issue type of thing.
That it was intentional or not.
I know you know Audio and video, but AI generation doesn't work the same way
Um... AI->Programatic Overlay->User.
Occam's Razor, to begin with -- less likely that there's a conspiracy to make songs sound worse.
More relevantly, I was part of alpha testing to try to isolate that sound 🤷♂️
Hello, i am using suno premium and have a problem with the model V4, if i want to replace a section of a song, i am unabel to change the modeln. the song was gernerated in version 3.5 please help
seems to be something limiting the length of song to 3:14 for me. Every generation ive made, even back in 3.5 was stuck at 3:14
Ok, so then you should understand. You said Audio, video, if you know AI then you know that Audio that is audible is nothing like generated Audio. Frequencies have to be controlled rather than Filtered during generation.
And the fact the devs were even asking testers for specific examples of the sound (which is why testers were trying to isolate it)
Call it my cynical nature. My first thought is watermark just because in my mind it's easy to do, and easier to mess up -- and even easier to openly use ai to make ai that messes up said ai.
If only -- debugging is primarily my day-to-day.
When you debug, do you think the bugs were intentionally put there by the developers? 😉
This isn't a developer, it's a business. So can you reword the question.
Nah, I'll just excuse myself from this matter -- today's been wild.
Do you think Suno would do this intentionally when Udio, one of their competitors, is actively developing a major release as well?
Selling a product with a watermark would decrease profits then...lol
Good talk. 🙂
I just got a crazy message when I refreshed. asking me to confirm that i'm using original lyrics to make music???
I think, I got that once. It was after the captcha thing happened
It’s a safety protocol to make sure that members are aware of the importance of using non-copyrighted lyrics. Happens to everyone
This suno's way of covering its ass?
I'm sure it is
never saw that before I guess they pushed it to everyone once v4 came out.
It is also the same message you had gotten when you first did Custom Mode. It appears now and the as a reminder (or cookie expirey?)
Maybe. I see it as a EULA ish thing
kk I was like "uhhh yeah,. I write all the lyrics."
I really wish there was markers when in Create mode. It is hard to identify where the AI is at in the prompt.
Look into EXIM frequency then do a little abstract thinking.
No the devs were specifically asking for this. I'm not guessing at this, this is what happened. And there was follow-up discussion regarding the issue. If it were intentional, they would say (like the time they did add an inaudible watermark and announced that.)
Sorry boss, I gave a quick response when I should have left it to the higher roles.
That’s a great idea, if you want to, please feel free to share this in #💡┃suggestions-web. the team checks that regularly looking for ways to improve
Right -- which is a reasonable thing to assume. Again -- I'm a little cynical when it comes to constants.
My premium account stops generating anything, cover feature not working properly, extend not working. I need Help!
i can not deactivate model V4
know what's crazy, I can't switch back to v.35
I can't even generate V4
for persona and the song text selection, the model v4 is always active
that may be the issue
does anyone have similar issues like me? I need help
i am using mozilla fire fox and i can use V4
with both chrome / firefox I can't use it
But it's was working before, i have 8 song generated with v4
ok its crazy
It's a serverside problem, nothing with the browser
@olive zenith It's not only V4, everything is not working, i keep recieving Credits Refunded1
For me 3.5 work
I created a new account, everything works perfectly, but i want my old library
i have about 20-30 section and 10 songs generated with v4
I doubt it, it didn't work weeks before v4 was launched
I contacted support team on email they didn't respond
Yeah/ that strange bugged Background Ticking 😦
https://suno.com/song/b6884081-8856-4c48-95a5-72bc2e3b8ec5
Song still cool though XD
circuitsoul song. Listen and make your own with Suno.
Can I get 5 free credits? I'm down to 5 but it won't let me make any persona songs without 10 in the bank, even though I have free persona songs stored up still.
I think it comes up the VERY first time you use custom mode, and never again unless you use a new device or the cache resets or something
yeah so far so good.
Do you have a trick to end music in V4 ? i don't want to waste to much credit on that
end music?
[end] Always works for me
[fade out]
it keeps going give me 2min instrumental and lyrics that are not there
I think i have enough credit
Interesting, I just figured out why the AI sometimes vocalizes random numbers. I think it is the token IDs in the prompt. Makes total sense now.
Hello. Can I run one of the songs I did thru V4 to clean it up?
you cn make a song if you have a subscrioption, it subtracts 5 from your daily
And is there way to download the vocals only so we can have stems?
are you on a subscription?
Not yet, I just started using this.
you can pull stems as a subscriber, but not as a free user
Awesome! What about running an older song I did thru V4 to clean it up?
you mean that sound in the background of the music that sounds like an old phone ringing? or text coming in?
again, subs for v4 as well right now
but there's plentty of issues with v4 right now
I understand. But can I run a v3 song thru v4 to clean it up without making any changes to the song?
remaster turns a v3.5 song into a v4 song
Wow, the 2nd one if so much cleaner!
except the major tickign inside it
OMG!
Yeah, that's weird. But the vocals are 100% better.
oh I agree, but the intrumentals are bad
Yes
Needs work
I'm sure they'll fix it soon.
I think most of that should be fixed for the public release (hopefully)
When's that? I do notice weird pops out of nowhere in v3
I have no idea, I'm not a staff member, just a user
Oh ok. In v4, will we be able to customize pitch correction to ensure it sounds natural and avoids robotic or autotuned effects?
A slider would be ideal to make it simple for all to use.
nope, nothing is different for us than in any other model
We need a way to customize the autotune because it will go nuts sometimes.
anyone notice the vocals are too far forawrd and the instrumentation is too far back in v4?
forward
I upload the songs I sing for pitch correction, harmonization and EQ. It's become my audio engineer.
PLus the music we can add is insane!
v4 covers seem broken to me, doesn't produce covers using the original melody
This laser beam warble sound is insufferable for v4. So strange.
Vocals are much cleaner but the timing is fked.
yea but in the mix the instrumentation is buried so far back
we can only wait for it to be fixed