💡┃suggestions-web

[potent crane]

which would be used by 1% of users, the vast majority have a post-suno app to edit and master, a free app like audacity will do more than you need.

[long bloom]

Well, I do use audacity, but that doesnt work if I want to share it via the Suno site.

I think that Suno should create a way to share finished, edited, AI-co-authored songs, perhaps with an "AI-coauthor" tag on a Jamendo or Soundcloud integration, so that AI Suno songs don't overwhelm the industry without tracking that it is AI and being filter-able.

I mean, It will happen regardless, but the problem can be mitigated with providing a easy channel to both share and appropriately tag creations

[glossy ginkgo]

Searchable library please

[hollow tree]

Suggestion to delete all songs

[glossy ginkgo]

Red "Nuke" button to clear library/created ?

[hollow tree]

Yea

Suggestion for a favorite section to only add songs one wants to save

so you don't clutter library

what do u mean by this btw

what is FurAffinity

so it has no relation to Suno

alright

[hexed spade]

A volume slider would be nice to have

[pulsar flume]

A simple question! How can I make money from it?

[devout sparrow]

When downloading music, those of us who work with it in our music DAWs often run into an issue regarding the quality of the tracks. It would be very useful if you could provide audio tracks separate from the songs, for example: one containing just the vocals, another just the drums, another just the bass, all in WAV format. This would allow us to work with higher quality, separating tracks into sound tracks in our digital studios.

[lofty wraith]

yeah i said the same thing a while back up there. its a must have .. for people that actually make music. without it. its almost useless lol

[karmic snow]

ability to save songs to be able to go back to them later and remix it (no limited page list essentially)

[rocky sedge]

I asked about this in another area, so figured I'd suggest it. For us who created songs in v2, perhaps offter a auto convert to v3. I've found that when i continue from the first part of a v2 song the v3 is very similar. But there is no way to recreate the first part in v3, so it ends up sounding kinda weird. So perhaps an auto convert to v3 would be possible since it seems to know how to go from v2 1st part and sound similar.

Also, I don't know how the software works, but getting it to create vocals and instruments seperately would be a life saver also, and I think would potentially equal higher quality output.

[swift totem]

I for one would like a search in the library for particular track names with also a filter too. I.e… I want to find a set of songs that I can remember part of the name Suno called it (‘no shoes’ let’s say) as me not putting in full name I search ‘shoes’, which is helpful to a point. But either need some form of genre selection.
I go first

[sage mirage]

Can we get the 'Continue from" time to stay put? Whenever you guys did the update to get the splash button that says "I won't steal lyrics" that time now changes when you toggle the instrumental switch. It used to hold the time chosen switching back and forth, now it defaults back to the full time and forgets what you set making it a pain if you are switching between instrumental/lyrics to try to get a good next part

[random heron]

The current quality of any AI music generator is not even near of normal audio quality. The stems would be as low quality too. Dataset training is probably 22kHz due the GPU limitations. But think it differently: It's great for vinyl style lo-fi samples. The higher quality comes in time, but we are not there yet.

Have anyone of you ever generated AI audio locally? With $2000 gpu it takes 5 minutes to generate 30sec of low quality mono audio. I'd quess it would take 30min to generate one 2min SunoV3 track. Low VRAM GPUs would probably not even run the model. Thats the biggest hold up for local AI audio and video generations.

[edgy grove]

^

[wooden fable]

@random heron i don't have access to the same algorithms, so i cannot compare. But for example "RVC" runs really fast here, even at 48 khz sample rate.

[dusk terrace]

Guys, you rock. You fixed this issue so quick 🙌🏼🙌🏼🙌🏼 Thank you so much

[wooden fable]

it's like half a second to replace the voice with another one for 17 seconds of speech.

so about 30x of real-time

on the other hand "facebook musicgen" runs super slow, and the results sound horribly bad.

[hollow locust]

Yes, I'm aware and that's not what I'm talking about. I'm talking about, for example, when the song fades to nothing, and there's a good few seconds left, then those last few, few seconds, the song starts up again. It ruins it.

And I'm talking about adjusting the length of the continuation. It's pre-set to one minute right now. Sometimes only an extra 10 seconds or so are needed to finish off the song... and it does finish it... before it starts adding more song... which brings me back to first point of needing a cut/delete tool.

Thank you. 🙂

[rancid pilot]

The explore page has no pagination feature? I saw a great song yesterday and marked it as Liked

I thought it will save to my Library and I just can't find it anymore 💀

And there's no Play History. Suno is killing me mentally by not having any chance to let me replay a song once I missed it 🥺

or Suno should change the URL when we played a song from the list, atleast we can have a Browse History from Web Browser for it

[astral dagger]

1, It would be good to have a dropdown menu where I can choose the possible styles. Generating the right music requires too much experimentation now, a style selector would be simpler.
2, It would be good if there were other options beside the instrumental menu. For example, if I'm a musician and don't need a complete song, just an instrument, then the system could generate one for me, be it a flute or solo violin. But it could also generate just vocals. I can currently separate the vocals from the music, but it's a cumbersome extra work, so this would be good.
3, I see that Suno AI uses the OpenAI Jukebox system, or if not that, the principle is the same. Therefore, I think it would be a reasonable suggestion to use more freely available unique vocal sounds, as the system can be trained on other sounds too. In most cases, the male-female vocal sounds are the same, with only minimal differences. There should be a vocal selection menu where if I can't choose the specific singer, at least I could specify the voice style. Baritone, soprano, deep bass, thin male voice, etc.
4, When generating a vocal version, with longer texts the lyrics inevitably slip, since the system divides the generation into blocks, and although the correct musical structure already works, this is not detectable in the generation of vocals. The singing regularly slips, and the lines of the bridge are already perceived as the chorus.

[rugged light]

There is the website audio-joiner, but I notice that the music loses some quality or volume. It would be perfect to have an editor on Suno

[rugged light]

I'm not sure if I understood your question, but if you want to finish a song, try writing "end music" in the style field. This has helped me to close the song in 10 to 20 seconds.

[somber laurel]

Is there a list of genres or instruments that suno V3 knows or have been discovered to work?

It doesn't know talking drum, which is one of the instrument I am trying to generate

[lofty wraith]

i make them sound amazing lol, its the parts where the bass drops and reverb comes in that the stems get ruined, freq of the other instruments cut the ai voice all up when extracting the stems. but if we could get a solo voice that didnt come in contact with everything we could for sure use the voices.

[quartz copper]

An option for download it in wav.

[karmic snow]

There NEEDS to be a feature where you can regenerate a specific part of a specific generated song without getting rid of the before and after parts

[rugged light]

I don't know if anyone has already made this suggestion, but it would be perfect if there were some fields to include:

→ Initial and chorus chords;
→ Instruments;
→ Interleave female and male voices;
→ Keep the same voice in all songs and change them, using some code to make them not similar to other creations;
→ Negative prompts, as someone mentioned here.

Currently, I include chords, bpm, genre, and voice in the music style field, but it is a limited field.

[lean island]

Can we have a little nummeric input here?

And maybe a (Last, (First - well refresh works) and a nummeric display of total pages?

[lean island]

Try doing an instructive "Meta" Tag instead [126 BPM] for example 😉

in the lyrics entry

is not always respected, but for me at least, like 50%

[rugged light]

Oh yes, that could be good as well, even though it takes up a small space in the lyrics

[lean island]

Most instructive tags actually are not worked down like for example [Intro] so that doesnt often not collide with the 8-lines suggestion

i do much mor than 8 lines on v3 (espewcially on the initial one where i place most instructions)

[lean island]

sent u some stuff as DM 😉

[worn atlas]

Just expermenting what you are saying. I have just put End in the style for the continuation . The song ends correctly than after few seconds suno went on singing again
#📚┃suno-school message

[worn atlas]

It doe not always work .

[hybrid crane]

I understand the interest in having an option for generating only toplines. As of now, Suno's AI music generation focuses on combining melodies with lyrics. While it doesn't offer an exclusive topline-only feature, you could experiment by minimizing the lyric input or using instrumental modes as a workaround.

[hybrid crane]

I understand the importance of having separate audio tracks for professional music production. Currently, Suno AI generates complete songs without the option to extract individual stems like vocals, drums, or bass. The tracks are provided in MP3 format. The Suno team is aware of the high demand for such functionality and is looking into it, aiming to deliver quality stems without instrument bleed, but this feature is not yet available.

[hybrid crane]

You can make playlist in the library now. 😉

[hybrid crane]

Your suggestion for an auto-convert feature from V2 to V3 in Suno highlights an interesting aspect of music generation continuity. Suno's V3 version does allow for continuing songs from any point and is backward compatible with V2. However, as V3 is still in the alpha development stage, there are certain nuances in how it handles song generation. It might interpret prompts differently than V2, potentially leading to variations in style, key signatures, BPM, and overall audio mixing and mastering. This could result in noticeable differences when transitioning from a song originally created in V2 to its continuation in V3. Until a more seamless integration is developed, users might need to experiment with different prompting techniques to achieve a smoother transition between the two versions​​.

[hybrid crane]

Popular by demand. Answered many times 🙂 See this answer: #💡┃suggestions-web message

[lean island]

The [Accapella] Tag might work... maybe add [Break] before when u didt work on something that u like

and make sure to write [Accapella] correct i have no clue how its written accurately 😄

[hybrid crane]

I understand your pain. I have it too sometimes. This is how I do it. I use the option to "continue from this song", where I want to end it (best is where it is already fading out), remove my lyric, and in the "style of music" I put in the word "end" without quotations. Usually it gives me a part with less than 10 seconds, so I get my credit back, and make a full song. That works pretty good for me.

[hybrid crane]

There is. In the right upper corner.

[rancid pilot]

I don't see the Pagination in Explore page but only Library. Tried in Chrome and Firefox

[hybrid crane]

Explore is not meant to have pagination

Explore is always the top 50 of users

[rancid pilot]

Yea, that's pretty sad, I'd like to explore/discover more interesting creations

🥺 Suno is now my new Spotify

[hybrid crane]

If you want to save songs. Next time save them in a playlist

[rancid pilot]

sure, lesson learned 🥲

[hybrid crane]

You can make playlist in your library since recently

[hybrid crane]

You have to look at it this way. Suno model understand everything like ChatGPT, but just is not always doing what you have in mind. That is still work in progress. But maybe the community wiki helps you a bit. See https://www.suno.wiki/

[hybrid crane]

You can trigger solo voices (semi) with adding "a cappella" in your "style of music"

[hybrid crane]

I understand the desire for higher quality audio, such as WAV format, for professional work. Presently, Suno V3 Alpha generates music at a 48000 Hz sample rate and 192 kbit/sec in MP3 format. This is a limitation acknowledged by Suno, as they work towards achieving studio-quality audio. It's important to note that converting from MP3 to WAV won't improve the quality, as the original generation's fidelity sets the ceiling. Suno suggests using minimalistic effects or simpler genres, like unplugged or hip-hop, to mitigate quality concerns in the current MP3 format​​.

[hybrid crane]

I understand your need for a feature that allows regenerating specific parts of a song without changing the entire composition. Suno is developing a "paint in" feature, which aims to offer just that. It would enable users to choose a part of the song they want to revise and regenerate it, maintaining the integrity of the rest of the track. This kind of precise editing control would be similar to what is available in podcasting and other audio editing software. Although this feature is still in the research phase and not available yet, it's anticipated to add significant flexibility and creative freedom to Suno's capabilities​​.

[quartz copper]

So it will come one day right? 🙂 Thanks 🙂

[hybrid crane]

Your suggestions for enhancing Suno's song creation process are quite insightful. Currently, Suno doesn't have the capability to specify initial and chorus chords, choose specific instruments, alternate between female and male voices in a song, keep the same voice across multiple songs, or use negative prompts. However, the concept of negative prompts has been discussed among the Suno team. This feature, if implemented, would allow creators to exclude certain elements (like specific instruments) from their songs. While these features are not available at the moment, your suggestions highlight important areas for potential development in Suno's AI music generation platform. Keeping track of Suno's updates might bring some of these capabilities to life in the future​​.

[karmic snow]

would this addition be something that is announced in #📢┃announcements ?

[hybrid crane]

In the future anything is possible, and the only way is up 🚀

[hybrid crane]

Fore sure!

[supple echo]

I second this graph view suggestion!

[karmic snow]

same

[alpine niche]

oh, that's why i tried many times to put "violin set" in prompt but always be piano. I also made a girl group k-pop song, and it works, i guess that's the advantage of the voice changing issue.

[karmic snow]

what about the ability to make a song with a single instrument? is that possible rn?

[hybrid crane]

I have plenty of songs only with piano, by using the instrumental option and adding in style of music "piano"

[karmic snow]

what if i want lyrics with it? im trying to make songs for characters that only use a single instrument

[alpine niche]

any other instrument only options?

[hybrid crane]

trial and error.. counts for last 2 questions 😂
There are no garantees..

I think guitar will do it too @alpine niche

[karmic snow]

well i was thinking of making the instrumentals first then "remix" that full song but if i remix it with lyrics it will just create a new beat and i dont want it to create anything new i just want it to apply lyrics to it

[hybrid crane]

I often start with an instrumental, listen where I want to start the lyrics, then fill in the right seconds (with continue option), and off she goes

[karmic snow]

well yea thats essentially what ive been doing but it doesnt achieve my goal if i ever wanna make a vocal version of a song ive done

[hybrid crane]

Do you have V3?

[karmic snow]

theres just not a way to do that that i know of

[karmic snow]

yes. i always use it

i only use the website

[hybrid crane]

This is a song I created today: https://app.suno.ai/song/be09fcce-5242-441f-93bb-60c94270a1f8

1. I went to create, clicked costum mode + instrumental. Filled in my genre and clicked create
2. Pick the best instrumental, remove the instrumental option
3. listen where you want to start the lyrics, click in the menu to "continue from this song"
4. Fill in the exact time where you want the intro to break of and start your lyrics
5. Fill in the lyrics, and click continue.

[alpine niche]

that's smart, i just now that we can start wherever we want by putting the time in prompt.

[hybrid crane]

Yes! it's great!

[alpine niche]

"hidden menu"

[hybrid crane]

Yeah website still needs work so things are more clear 🙂

[turbid yoke]

I would like to suggest keeping the song of K-pop tutorial. If it's not possible, can you please share that song with me, please. I really liked it. 🥺

[hybrid crane]

I am not familiar with that song. I personally have no clue 🤷‍♀️ 😦

[turbid yoke]

It was in the explore section about a month ago. The lyric of the song was a tutorial about how to make a song in Suno. 😢

[hybrid crane]

that's like searching a needle in a hay stack... I theory it should also be in #🐣┃suno-showcase somewhere. If you know (part) of the name you might be able to find it.

[woven cradle]

If you know the time period you heard it, try searching your browser history.

[ember mountain]

I think a good Change would be that suno doesnt Cut of Songs somewhere and gets more cleaner

[lean island]

i could imagine an upgrade to 320kpbs (from my point of view), but a wav training dateset for all genres, uff... hard to imagine. What i would like: more love for the bad genres, not that im intrested in, but those who are - it's a bit unfair - when i compare results from my style with their... thats not just a gap, thats like a decade gap of style and an extreme one in terms of quality.

[ember mountain]

Im interested in more Love for Nu Metal and all Rock Genres

[lean island]

what i personally would love: unmastered output, but thats like wav quality something i that will probably keep a dream for long time 😉

in a perfect world we would get every single mixer track from the unmastered song ^^

[hybrid crane]

https://tenor.com/view/it's-just-a-matter-of-time-doctor-cardosa-marco-pigossi-gen-v-it's-gonna-happen-either-way-gif-16153807130615846716

[lean island]

😉

had a glimpse on some new architectures, not music, but heck - the steps are those of giants... reminds me of slightly of the doubling the mhz every cycle times 😉 and we're still on the begining of the era of AI.

so yes, youre absolutely right @hybrid crane

@hybrid crane how are chanes for slightly larger style prompt btz? ^^

*chances

[delicate cedar]

Please read the #📜┃rules particularly #3️⃣, thanks.

[topaz bluff]

Good evening. Can you tell me if you plan to add the Ukrainian language? https://github.com/suno-ai/bark

[lean island]

i think bark is and stays what it is in terms of features - correct me if wrong - you could modify itself if you have the knowledge and skills.

[hybrid crane]

Don't repost your questions in several channels, please

And as I said: Suno is currently more focused on developing AI for music creation and is no longer actively working on the Bark platform for speech. While Bark is not operational on Discord at the moment, you can still access and use it online here, and you have the option to install it yourself. The original version of Bark is available on GitHub, along with a downloadable GUI here.

[lean island]

Make sure to check the forks, maybe what u look for somebody implemented already 😉

on github i mean

[raw cave]

Didn't I? VOICE CONTROL.

[hollow lion]

Can you please remove Cantonese Chinese from the AI? Currently, every time I create a Chinese song, I waste at least 500 credits. It has nothing to do with the melody, but Mandarin Chinese and Cantonese Chinese are mixed together, so the whole song cannot be used, even if the melody is good.

[hybrid crane]

I understand the frustration you're experiencing with the AI blending Mandarin and Cantonese in your music creations. At present, Suno's AI doesn't possess the functionality to isolate and exclude specific dialects, such as Cantonese. The technology's ability to precisely manage distinct dialects or languages in music generation is still a work in progress. This issue isn't unique to your situation; for example, as a Dutch speaker, I've encountered similar challenges with songs adopting a Belgian dialect. It's important to recognize that AI in music, being a relatively new technology, is continuously undergoing development and improvement. As it evolves, we can expect more refined control over language nuances in song generation.

[hollow lion]

It will waste so many credit. I try to add mandarin on style, but also mxi cantonese

However, I appreciate your project and your efforts

[hybrid crane]

If (as in your case) the specific language prompts are leading to the excessive consumption of credits without the desired results, you might want to consider using more general language options, such as English, which the AI may handle more predictably. This approach can help you conserve credits while still creating music until the technology further evolves to better differentiate and handle distinct language dialects like Mandarin and Cantonese. 🙂

[lean island]

Maybe do it in english, check if any voice changing AI supports what you need and switch Vox? Or remove them and - as last possibility - use it and hire a session singer? Might be worth a try.

U could try to add [Muted Verse] and let i make the song continue without the verse actually hearable to make adding vox easier afterwards and not requiring another paid service.

but still keeping the influence of the words mostly 😉

as with many tags, it's not always respected - but i usually get what i need within a few tries - unless i navigated the creation process into what i call "nearly dead end" where u cant make it do what u want really 😄

You know you can make it sing "La La La or Mmmmm Or Ummmm" no real lyrics, just sounds made 😉 But if you want to keep a verse or chorus, but have no lyrics - try "[Muted Verse]" like i suggested in my previous answer

[dusky lake]

i imagine something like this would be very far off... but i think it would be useful to have the song export as instrument and vocal tracks... so you could do your own mixing if you wanted... i find that in general i would wind up tweaking audio levels and such a bit if i had that the option to do so

[hybrid crane]

I answer daily questions and suggestions like this. They are in high demand. See :#💡┃suggestions-web message 🙂

[lean island]

That would require a completely different approach AIwise... model-wise, training-wise, and getting that training data for songs - so you have everything seperated, isnt easy - you need - for every genre - data examples in quantities you barely can't imagine yourself if you never did it before, to reach a such a quality...

[frank spruce]

Thank you very much. I will try!

[lean island]

But one day, that will be possible too - AI changes boundaries. Its just not something that i would expect anytime really soon...

but if you just want the vox for examaple, there are AI models that can extract vocals more or less "acceptable quality" 😉 depending how good the recording is and how well the vocals are clearly seperated from the music frequencywise..

[bleak scaffold]

Suggestion 💡: Make V4 model with clear voice for Alternative & Punk Rock songs. Make it 10 times the price of V3 model (to cover the cost of increased computation). We will use it with pleasure ❤️ Currently, when the sound isn't loud - the voice sounds crisp & clear, but during Rock chorus it's too hard to listen to it due to voice distortion 😩

[lean island]

How are the results on those genres beside the vox issues? Good quality?

[bleak scaffold]

Besides the voice it's good. Sometimes there is a bit of clapping on loud noises, but overall it's good (instruments are good, but voice is so so)

[lean island]

that "clipping" or "distortion" is a known issue on v3

it sounds like clipping on a single instrument or vox or maybe 2, but not on all elements

[bleak scaffold]

I tried to separate voice & instruments and fix voice on chorus, but it's a bit complicated to separate instruments and voice during loud chorus cause they are almost blended into one. Still possible to separate, but some instruments might be picked up as well

[hybrid crane]

Regarding the suggestion for a specialized V4 model for clear voice in Alternative & Punk Rock songs, currently, Suno's development is focused on the existing versions, with V3 being the latest and still in Alpha I might add.

Suno's AI, as it exists, strives to balance quality, versatility, and accessibility. While the idea of a specialized model for better voice clarity in louder rock segments is intriguing, it might be more feasible to refine the existing models to handle such scenarios better. Suno continually evolves based on user feedback and technological advancements. So, while a specific V4 model for your described needs isn't currently in development, Suno values such feedback and may consider it for future improvements.

For now, to minimize voice distortion in loud sections, you might experiment with different styles, metatags, and the balance between instruments and vocals in your prompts to achieve the best possible output with the existing models.

[languid phoenix]

try some of these styles:
https://app.suno.ai/playlist/a10c52fd-647d-4bd3-96ca-a5c9308250e6

[bleak scaffold]

Thank you! I'm experimenting 🙂 Just wanted to brainstorm some solutions. Improving existing models is definitely a great idea.

[hybrid crane]

The only ways is up 🚀

[dusky lake]

yeah i imagined this would not come for a while... but to your point, Ai changes the the boundaries... not only that, it also changes the game... so it becomes more of a cost vs benefit equation... instead of "is this possible!?"

right now, i imagine this is well out of scope, however fast forward a few years... maybe not so much

[hybrid crane]

here an answer from a DEV about this: #💡┃suggestions-web message

[dusky lake]

thanks for that @hybrid crane

[lean island]

@languid phoenixYou got it 😉

[lean island]

i got that with my genres too... using a single genre or only 2 doesnt yield the results i expect, adding two other related genres into tag increases likelyness of getting the style i want - and dramatically increases quality compared to using just the genre i actually want 😉

[languid phoenix]

🤖 💬 Yup!

[lean island]

and important to make it one tag, it may be conincidence but it feels, if i seperate with coma for example, its not doing it the way i want it 😉

[languid phoenix]

ya i rarely use commas in my Music Style codes

[lean island]

hard to really say as there is no reproduceability

[lean island]

so you feel it too? its not just me 😄

[hybrid crane]

I want to remind you guys this is the "suggestion" channel. For free chat please go to #💬┃general-chat

[rain kelp]

suggestion: A way to invest in suno

[hybrid crane]

Getting the premium plan is a simpel way.. 😉

[vast bough]

❤️ SUGGESTION: I'd like to be able to cut or trim clips/songs because sometimes clips/songs are prefect except for the beginning or end. ❤️

I waste a lot of credits because the guitar comes in suddenly 10sec after the song ends

[hybrid crane]

If you have a paid sub you can easily fix this. I exactly explained it today earlier here: #🪲┃bugs-web message

[weak trail]

+1 for Paypal

[rotund orbit]

ooooh, the continue from field is back. well color me happy

[hexed minnow]

I'm sure this suggestion has already been mentioned. I propose giving "Style of Music" in the Custom setting a weight, similar to the Image Weight in Midjourney. This would only apply to songs that have been continued. You can get some really interesting results by mixing styles after continuing.

[hybrid crane]

While the specific feature of weighting the "Style of Music" isn't currently implemented in Suno, a similar effect can be achieved by repeating metatags or genres in your input, which gives them more weight in the song generation process.

[hexed minnow]

Is it inappropriate to say that I love you? Awesome information! Thank you so much!

[rotund orbit]

one suggestion: I want a box for chord progressions longer than 120 characters. Now that we know the machine takes them and reads them, mostly correctly, enough of the time to have it work

[lean island]

i would suggest linking the community wiki directly in the welcome message as its really helpful, would probably avoid a lot of repeated questions.

[half sail]

Some quick suggestions for the UI and Website:

1. Some kind of visual indicator that i have or have not listened to a certain generated song would be nice.
2. The ability to list all "Full Songs" in my library.
3. The ability to have all prior segments of a song listed under "Continue from" as a selectable options
4. The ability to setup your own style configurations and save them as a dropdown option in the "Style of Music" field.
5. Auto generation of a Song title by AI if left blank by user
6. The ability to readjust my song and lyrics preview on the right side (adjust width by default or user setting)

[tepid trench]

Length threshold for automatic credit refunds seems unreasonably short and not intuitive/obvious. An 8 second generation did not get refunded, previously they have, but it seems it depends on whether it's exactly 8 seconds or slightly longer. 10 seconds would be a more obvious length (maybe anything under 11, so as to accommodate e.g. 10.9 seconds)

[delicate rain]

I would like to subscribe but don’t have a creditcard 😢 is there something else possible?

[turbid yoke]

I'm wondering if you are considering an API service. If that's the case, please consider a Pay-as-you-go option.

[tardy pivot]

I just lost about 2000 credits because the credit tag was not renewed by my upgrade to Premier. That's completely stupid. In the worst case, I have about 24 hours to spend 10000. That's quite unfair and not really thought through. Maybe not even feasible. And then this (that the date doesn't change) is not even communicated. It was a mistake to upgrade... Where can I talk to someone about the above, there is no support channel.

[formal bear]

email [e-mail]

[placid flume]

**Please add Digeridoos!! **

it doesnt work even when it is the only prompt

I need my trance mix to have a digeridoo solo

[placid flume]

https://tenor.com/view/letterkenny-didgeridoo-music-rip-ripping-gif-20320406

[somber laurel]

Can we have a search for clip function maybe via style of genre entered because it's getting hard finding clips create months ago especially if you have been paying for pro plan for a while

[worn pelican]

I am sure this suggestion has been mentioned, but having a search box in the library could be helpful. Like it would be nice if I could search for all tracks I've generated that contain the word "rock", either in the title, prompt, or lyrics

I would imagine this feature is already being worked on, but thought I'd add a +1.

As a side note I find it mildly amusing that an advanced AI company doesn't have a search feature in the library yet. But I know that day well come. Nice job with your V3 model I'm really liking it so far.

[brittle mist]

There was an option to see which part was previously used when select a track to easy find the parts used on a final song (and ease delete all part we don't use), why it dessepear ?

[gusty latch]

Also idk if it's been suggested yet but can we get a skip to page option for the library?

[gusty latch]

Also, would be nice if you put the creator's name tag on Public posted songs. At least the ones who are subbed when they Public it.

[languid kestrel]

It'd be a huge, immense problem solver if we could somehow predetermine a minimum length for a generated segment. I can't tell you how many times I've been forced to try and generate a new section because a clip cut off 3 seconds before it finished a chorus despite being at less than a minute in length. I've yet to find a solution, sometimes Suno just decides "screw you you aren't allowed to finish this song properly despite getting everything perfect no matter what you do" that day and it's several hundred credits in the hole.

[desert whale]

We’d would absolutely love the ability to export as wav. Likewise it would be amazing if we were able to export as stems. I know that’s a big ask but I’m optimistic it’ll eventually be a reality?

[desert whale]

Yes I feel that frustration but also on the flip side you can already create pretty amazing compositions with enough trial and error human ingenuity. I’d prefer the team focus on upping the audio quality so it’s less choppy, and prioritise higher resolution output and in a perfect world an export as stems option

[brittle mist]

If you can add a shortcode like [guitar solo] or [backing track] thanks

[spiral acorn]

Will we be able to search in the Explore tab of the website? I want to find the tracks I like best, and look at how they construct their lyrics / song description etc to help me learn what achieves different sounds

[earnest leaf]

Please make it so that when a prompt is flagged for moderation, it highlights what phrase triggered it. Gets very annoying writing up a whole set of lyrics to try and guess which ones the ai deemed offensive.

[trail spear]

why there is no paypal as a payment?

[lethal sphinx]

Asking the same question as the guy above me.
I want to give you money. Please give me a way to do that that does not involve credit cards -- which aren't particularly common in my country.

[cobalt pecan]

Allow us to change the displayed lyrics on a "Full Song" render, as they can be entirely inaccurate after using the continue function if you are cutting tracks short on continuations (and not just continuing from the end each time).

I just finished a song and because it had several renders (I only used about 20s each from multiple tracks, using the continuation function), there's just a huge wall of text for the lyrics that are all repeats or simply aren't in the song at all.

The original prompts would ideally be shown in another tab.

There are a few places the lyrics are wrong after using the continuation function multiple times: The Music video, Lyrics side panel while creating, and the dedicated page for the song when clicked.

Example: #🐣┃suno-showcase message

[lofty sonnet]

Please Smooth white noise and transitional fx's, sometimes in genres where that stuff isn't as modulated, white noise sweeps can gate, duck from compression or tonality shift strangely

Search function in Library

[hidden lagoon]

Please add the option to export the channel, or at least the instrumental / vocals alone.

[lofty sonnet]

Please add the ability to find a V1 from a V2, etc because I sometimes come back to working on a track and have difficulty finding it's Part 1

could be on the right when you select a track you can click an track art image drop down into its previous versions to be selected as track of choice on right, then V1's display entire tree of non-deleted gens

idk, gen tree suggestion again

Search function could solve the recovery of V1 though, unless title was changed and then FORGOTTEN, but we want tree visual and interactivity, ability to rearrage of even make prior to ___ gens (previous parts) continue from negative -02:00 or -01:00

[lofty sonnet]

re-gen of limited and taggable layers

ability to Modify this song, re-gen/create duplicate of exact song with: (delay on vocals, livelier drums, a funk rhythm guitar layer, add background party ambience) without: drums, any subwoofer content, cymbals, fx, reverb, that entire plucky melodic synth layer... for stem creation and removal of elements with clarity and pre-multiband/sidechain compression

[rugged relic]

^ ban this user for lewd profile picture

[eternal sparrow]

Let subscribers buy credits at an appropriate price. If we sub for 30$ for 10,000 credits, it'd be nice to be able to buy 10,000 more credits for another 30, instead of 4,000 for 30.

[delicate cedar]

Per rule #2️⃣, this is an official mod request that you change this profile image to conform to the policies on this Discord.

[rocky trail]

Chirp:

1. More characters to get into deeper detail on the mood/genre/style of the song. Get more atmospheric detail and song sequences etc.

2. Buttons to select a Male/Female/Robotic etc Voices.

3. Allow to possibly change in which order the parts come in. In example: The 2nd part actually comes 1st rather than the Original 1st Clip.

4. Allow us to change the sequencing and display of the lyrics because the song sometimes skips over lyrics or changes the order.

Love to see this in future! It’s been Great

[delicate cedar]

Thank you for bringing this to our attention, however there is a nicer way to do this and it doesn't involve telling moderators what to do. Next time please simply ping one of us and let us know about it or send a DM regarding the situation. Thanks.

[leaden marlin]

Deberia poderse descargar la. Base instrumental creada por suno en pro y Premier.

[soft oasis]

Maybe you should make an evolution of suno video. Show seamless transitions from V1 to V2 to V3

See how suno has evolved

[gusty latch]

Not sure if it was already said, but keep the Version choices for the future and include all the older and newer choices for subs. ❤️

[soft oasis]

It's seriously has become so good

[summer hollow]

would be a good idea to be able to save voices so if whe want to use the same voice again whe just need to select it suno dosent understand what kind of vibe is the song exemple sad depressing that would be interesting to add

[hybrid crane]

The AI generates voices and music dynamically each time, based on the input provided. This means it doesn't have the capability to 'remember' or replicate a specific voice for later use. To achieve a certain vocal style or mood, your best approach is to give detailed and descriptive prompts, guiding the AI as closely as possible to your desired outcome.

[summer hollow]

yea but here is no possiblility to give a prompt like that when using the custom lyric function is there ?

[hybrid crane]

you prompt in the style of music, there you can say for instance "sultry voice". And you can do it in the lyrics [smooth voice] for example....

[summer hollow]

mh is there a way to make it the voice is a male voice or some places there are high ntes

[kindred carbon]

I have a tip/process for the community that I feel is valuable: Most of you that are new to music production are getting a final result fast and then having to deconstruct it re how to edit/build on it.. So, Instead of relying on Suno as a full production suite, most are going to make their way to getting a DAW, and stemming these out into proper arrangements there vs relying on Suno for the whole editing process.

I have all the best separation tools, & have recently found a free one that allows you to test all kinds of AI models including newest ones not available in any pro tools yet. I advise you all to get UVR5, it's free & has amazing models to separate the vocals from the instrumental at least, even backing tracks/harmonies from main vocal (more expensive tools will give you all layers, RIPX & SL10 being leaders (AND LALAL . com is the best web-based isolator as it does processing to the rip that normalizes & deverbs it, & is cheap/has free options).

Attached is a screenshot of UVR5, maxed out, took 13min (because I'm using the newest/most refined model (MDX23C). Other models are under 10seconds and faster on an Nvidia card).

Just google UVR5, it goes to their GitHub, & get your DAW of choice, most have free trials, Audacity has layering (free), Mac has Garage Band already installed free. You'll love the control you have, tying parts, cutting/extending/intro/outro/moving words/lines, put pieces into proper arrangements that you can then master, fx, use VST's, prepare commercial releases, etc.

Until Suno can spit out .wav stems generated that way vs an mp3 being upsampled, this is a good solution.

Lastly, you can use a variety of voice changers/cloners to capture a voice/male/fem and stick to it. I use Vocaloid Changer, and a few web-based ones (Lalals - different from LALAL), that way you can retain any voice. Strip vocal > Send to Lalals pick voice > export back. Different tempos or instrumental changes, the DAW can fix all that quickly too.

Bang bang.

[rigid ibex]

Hey does anyone have any song ideas I have a guitar and piano I have ran out of ideas

[unborn token]

i feel its very very hard now to get V3 to do an instrumental inside a song. usually it only took [guitar solo]. now it just ignores those commands and free styles the chorus again or the last verse instead. really noticeable change

[hasty pagoda]

In the future do you think suno will be able to take audio uploads, so you can upload your audio file, and the suno ai will create a backing track for the song your singing in the audio file ??

[sage mirage]

Probably not. It's most certainly capable, but it's a copyright issue. There isn't a way for Suno, or anyone AT suno, to make sure you own the rights to whatever you are uploading

Audio is audio is audio

[hasty pagoda]

ahh i see

[sage mirage]

Yeah

Now if you could somehow use audio you generated within Suno to inform new generations I think that would be sweet, a way to keep styles or maybe even singers, and since it's all within Suno you get to skirt around the rights issues

[hasty pagoda]

Well i guess we can kinda do it the backwards way atm then. So maybe I put my lyrics into suno, have it create a song until i find a backing track i like, and then seperate the lyrics from the audio somehow

[sage mirage]

Yeah

[hasty pagoda]

i just dunno how you seperate audio from the backingtrack

ill try have a look online

[sage mirage]

Just a bit up in this chat musicalony posted a thing

[hasty pagoda]

oh nvm, the guy above just posted one 😂

[sage mirage]

It's seriously like

[hasty pagoda]

Yeah just saw XD

[sage mirage]

yeah

[hasty pagoda]

wtf are the chances 😂

[sage mirage]

I haven't messed with anything like that but hey: Lots of people want the same thing you do

control

[hasty pagoda]

Yesssss

love control

Ill try that and see how it goes

[rugged light]

I would like to suggest the option to select the songs and decide what to do with them. For example, delete some or all of them at once while they are in the trash 🙏

[worn pelican]

I can suggest an amazing feature. I am actually shocked this doesn't already exist. So how about when people select multiple songs, they can download all of them? Literally not fun going through song after song, downloading them individually.

You can add multi-selects to playlists...but not download..

Still have my fingers crossed on sorting/searching/filtering our libraries. That will be a glorious day.

[rugged light]

One annoying thing that happens in the app is that some texts turn black, including the song lyrics, as you can see in the screenshot. It's not every day, but it's annoying when editing the song lyrics.

[junior oar]

pon esasugerencia en ingles

[hexed minnow]

I'm sure this was already suggested, but just in case... When "continuing from", it would be VERY helpful if the timestamp could be listed in the metadata somewhere. I experiment on where to continue from and after generating a few I can't tell where where the timestamp I started from is. Thank you so much!

[gusty ibex]

They should put the option to explore a search engine to be able to search for songs from different artists who have created their music.

[full heron]

My idea is an Option to add your own voice.

[viscid temple]

would be good to be able to approve the chatgpt lyrics before chirp starts generating the whole thing, just in case you don't think they're appropriate

[sharp meadow]

Think other people might have mentioned this but it'd be super sick to see an ability to upload an mp3 of a song you love the most for suno to analyze the music style of, then allow you to tweak the details on what it gives you back. Since I find that trying to get a song you want to make in the AI sometimes is really difficult to get to a sound that's similar to other songs you like, out of your prompt.

Like if I'm wanting to get a Madeon or a Quest For Fire Skrillex sound, its decently hard to unless I get close to that sound design in the prompt while I use up a lot of my credits to regenerate and hopefully find something I like.

[formal bear]

you can do that in the not one box mode, just click generate lyrics, and edit...

[viscid temple]

what is the not one box mode?

[feral aspen]

Custom Mode

[near orbit]

I would like to ask, if you purchase a paid membership, will the music production be extended from the specific 1:20 to more than 4 minutes?

[leaden marlin]

📌 Suggestion: Improve the posibility of listening the public songs that other users are creating/sharing as a main feature of Suno.

• [ ] Explore--Search public songs by language (add lang select box to Custom Mode)

[shrewd zinc]

Hi everyone 😃 I wanted to ask does anyone with paid membership know how to download wav file instead of MP3 ?

[sage mirage]

It's only available as an mp3, there is no .wav

[shrewd zinc]

Do you think they'll bring wav in the future? and also do you have any trick to download the exact instrument that was used ?

[sage mirage]

Suno doesn't make music like you think it does, Suno directly outputs the complete mix

The lyrics/voice and music are like, one lump

You can use other tools to break up clips from Suno

[shrewd zinc]

Ah now it makes sense, I really appreciate 🙏🏾 one last question how do you generate the full song without having it just cut off while playing the song? and is there a trick to make the song fade in and fade ou?

so it's smooth

[sage mirage]

You know you can continue a clip, right?

Suno is an AI system so it's weird and does strange things sometimes

[shrewd zinc]

yes but it doesn’t end smooth

[sage mirage]

Right so really the only way to do that is to keep trying. Cutting off and generating from different points in the clip you are continuing from

[shrewd zinc]

I wanted to try commands and see, because I noticed under the Song Description section if you say the type of artist whether asian or black american from NY it gives you a very good referencing, i'll try to say fade out at the end smoothly and see what happens

[sage mirage]

Yeah, you can also spam things. I have had decent luck making an entire new clip just for the ending, and changing the style of music to 'END' spammed a whole bunch

[shrewd zinc]

Can you elaborate coz i'm lost a bit

[sage mirage]

If you are wanting it to end smoothly try doing this as the style of music for that ending clip

[shrewd zinc]

Ah okay i'll try that now

[sage mirage]

Nothing is 100% with AI models like Suno

[leaden marlin]

Kaker: is there any prompt for just voice, read a poem or a haiku? (or a minimal classical music as background)

[sage mirage]

Man yeah so there was someone that was making PA announcements, like you'd hear in the 80s in a store

Something like [Spoken Word] and then whatever you want them to say

You can quite literally ask it for exactly that

For the Style of Music put something like "spoken word, haiku"

See what you get

[leaden marlin]

Thanks, i'll try

[sage mirage]

You can then completely change the clips you continue on from that

There is no reason you can't start with a spoken word poem then drop it into a country western love ballad, why not

[leaden marlin]

This worked: deep male voice, spanish, spoken word

[sage mirage]

Rad

[hollow locust]

Add Tuvan/Mongolian/Tibetan throat singing to the training data.

[cyan snow]

Hope you're all doing well. Here's a few things.

1. When creating a traditional pop, rock, rnb, rap, country etc.. song, the AI tends to forgot the exact chorus sound and sometimes even forgets it 100%. This happens due to having to generate smaller clips after the first one, and especially if one needs to continue from a point within verse 2 or pre-chorus, which leads to it forgetting parts 1-2-3 of the song generated before that generation.

2. When trying to have a short intro before the song starts (e.g. 5-12 seconds), many times it doesn't even recognize the idea that an intro was requested, and sometimes it recognizes the intro (piano, electric guitars, etc..) and decides to make it the main part or the only part.

3. Many times there is more time in part 1 to complete more words (e.g. rap or a pop songs) than what one can place inside the box, this leads to short part 1 (less than a minute sometimes), and as I mentioned in issue 1, that leads many times to terrible future parts structure. What also leads to issue 1 is when one can place verse 1, pre-chorus, chorus, verse 2, (pre-chorus 2), but cannot place chorus 2, which many times leads to a very different chorus 2 (and for consistency many/most songs want a similar chorus throughout the song) because Suno forgets.

4. When trying to end a song and one places a fade or an end, it does comply, for the most part. Sometimes it does it even at the cost of cutting the song words written. It is even more evident when in the final part one tries to place two choruses, one being final chorus, for the most part it skips one of them or does a merge that is not what was needed or wanted.

Since this is the suggestions channel, here are a few suggestions to fix these issues:

1. An obvious solution would be the ability to create a full-length song 3-4 minutes, or somehow having the [chorus] tag remembered from earlier parts even if the part it's following did not have a [chorus].

2. Intros should be easier to define even by seconds [Intro Emotional Piano <12 {seconds}] or [Intro Emotional Piano >5 <14 {seconds}].

3. The box to place words should have the possibility to place the whole song. this may also help the AI understand the song as structured. An optional thing would be to allow the user to input their preferred song overall time within a half minute frame (e.g 3:00 or 3:30 minutes) and Suno would pace accordingly.

4. Fading and other styles of endings should only be attempted after song is complete and not just to fade/end by skipping parts. If there is a double chorus they should not cancel each other or be merged, they should be read as separate entities with the possible chorus or final chorus sung with slight changes as is done in many mainstream songs.

Thank you for your time. Love the product. Would love to hear from you about these issues/suggestions. Have a wonderful day/night.

[pine tree]

Would it be easier to find library files if there was a search function for lyrics? Would it be convenient?

Vocal only track "Music genre name".

Vocal track, accompaniment track, drum track.

Drum solo, guitar solo, piano solo, bass solo.

If the number of vocalists can be set, hamori and back chorus will be added.

It is good if you can decide the direction of the development of the music.

It would be better if the name of the music genre and performance style could be freely changed in the middle of the song.

[nimble island]

is there an api?

[hybrid crane]

I understand your interest in having Suno analyze an MP3 of a song you love to replicate its style. But Suno aims to produce tunes with original vocals and encourages creativity rather than mimicking popular artists or songs. This approach is not only about respecting copyright laws but also about encouraging users to explore and express their own creativity. Suno's goal is to empower users to create new music and lyrics, fostering a unique and individual musical journey ​​.

[hybrid crane]

While a paid Suno membership gives you access to V3 Alpha with up to two-minute song clips, for longer songs, you can use the 'continue' option. This feature allows you to generate additional parts of the song, extending the total length beyond two minutes.

[hybrid crane]

The suggestion of enabling Suno to create longer songs at once poses several challenges. Firstly, longer song generation demands more computing power, which could result in higher credit costs for users. Additionally, the current Suno model is still in its alpha phase, meaning it's undergoing continuous learning and improvement. At this stage, generating very long songs might lead to more frequent instances where the output doesn't align with user expectations, resulting in wasted resources and efforts. It's crucial to remember that AI models, especially those in developmental stages like Suno's, require intricate and complex learning algorithms. Enhancing their capabilities goes beyond basic coding changes and involves a careful balance of user expectations, technical feasibility, and resource management.

[cyan snow]

I appreciate your response, you are always very helpful.

Personally, this may be just me, but if I'm spending the credits anyway (and sometimes way more because of many bad outcomes due to what I wrote) I'd rather have a four minute song taking 10-20 credits than a song broken into two or seventeen parts which does not give me anything near the result I need or want. There can be a box with the ability to have an x minute song (e.g. 5 minutes) and it would mention that it'll cost x (e.g. 20) credits.

If you look at the #🗳┃mini-polls message [mini poll] about ideal length the overwhelming desire is 2-4 minutes (>70%) and if you add those who want 4-6 minutes (<86%) it's one of the main issues of producing a great final song, instead of a few attached parts, which may or may not stay the course. For me the ideal length is between 3:30-4:00 for most songs.

I wrote these suggestions, not because I thought there's an overnight fix. Rather, I thought to explain a few things that I gathered while using Suno for the past month, which would enhance the overall experience of Suno. I hope the team looks at them.

Thank you.

[timber halo]

^ this please! Just standard social media and forum user / content click and search features please. Right now, it's impossible to find other creations by the same user afaik and that would be super useful for the community and learning prompts etc.

[peak pasture]

Plz add Paypal because the most European dont have/use Credit Cards!.

[arctic ibex]

Please publish guidelines that explain why certain prompts are sent to moderation/refuse to generate. I suggest breaking them down by possible errors, so that users can troubleshoot on their own.

[hybrid crane]

The preference for PayPal, particularly in Europe, is understandable given its ease of use and widespread acceptance. However, Suno's decision to not offer PayPal is tied to their payment processing through Stripe, which does not support PayPal as a payment method in Europe. I myself am from the Netherelands, and I use a prepaid credit card. These cards work similarly to regular credit cards but offer more control and security, as you can load them with a specific amount of money. This way, you have a flexible and safe option for online purchases and subscriptions, including services like Suno.

[worn atlas]

Spending more credits for longer song generation makes sens for me too. Finally the user will have less trials and less crédits consumption in continuation mode. It looks like a win-win mode for suno and users.
This has been pretty well suggested here
#💡┃suggestions-web message

[rotund orbit]

thought, one of the biggest problems I've been having is getting otherwise perfect takes that don't have musical intros. Now, I like the change to continue from, having that field, essentially trimming and appending my track to make the next piece of the song. And that helps get rid of the weird repitition thing that v3 is doing. But, what if you did basically the same thing, but instead of appending, you're prepending? That would make it easier to do live tracks and musical intros

[swift parcel]

Personally, I'd like to see the song info show the time it was created in addition to the date it was created.

I would also like to see a page listing the playlists and the ability to sort within the playlists by name, time created, etc.

[woven pilot]

can we have volume control on previews?

[sage mirage]

Seems relatively easy to add vs the other suggestions usually posted in here. I'd appreciate it too.

[rustic tapir]

A volume slider would be nice! So that i dont have to change desktop master volume everytime

[sage mirage]

I'd also like the ability to pick any image generated earlier in the song and use that as it's main. Lots of times I use a whole generation at the end for the ending, which usually means it ends up as a bird because that's the default image. I'd got plenty of images that work really well earlier in the song but Suno defaults to the last one which means the lyrics never get used for the image. I like birds but I don't want just birds, you know?

[potent crane]

Most Europeans have both, to be fair

[fiery linden]

Hi, I have some suggestions for future features I would like to see:

1. The ability to clone a voice that's generated by a Suno AI output. No uploading voice from outside source to avoid copyright issues. The reason why I want this feature is so I can have consistent voice for my virtual singers. Right now I have to spend so many credits until the voice somewhat matches the previous voice I like.

So here's how I imagine you implement this feature. When you create a new song, you add an option that lets you use another song that you already generated as reference. But the reference is only for the vocal - not the music. It's similar to how you continue a song right now but this feature would work for a new song - not continuation.

2. Download music and vocal tracks separately. Not sure if it's possible but this would help with editing the music or the vocal track.

3. Right now with 3.0, a lot of times the AI doesn't finish the last word. It often cuts off before the last word and it's difficult to piece them together with continuation. I know it's alpha right now but I hope you guys are aware of this issue and will improve 3.0 so the AI will make sure to always finish the last word.

[rotund orbit]

one easy quality of life improvement I would like to see is the audio sampled at 44 instead of 48. It would be fewer export steps

[glossy ginkgo]

Suggestion: some music related emojis for this discord server

Like an animated equalizer

etc

[shadow sigil]

I have some User Interface suggestions, I hope and think they probably were mentioned, but just in case:

1. I'm missing a search function in the Clips Library.
2. I'm missing a: Multi-select + Multi-select delete tool for clips.
3. The option to add tags to certain clips and songs, for easier search functionality

[hybrid crane]

Yes all are already mentioned.. Let's just see it like this.. Great minds think alike 🙂

[shadow sigil]

Oh thank god. Fantastic, also any news on when any update is made/is a work in progress? Like a update page or channel? (First 10 minutes here in this channel and relatively new to discord)

[hybrid crane]

check all the channels created.. there are a bunch.. but #🚀┃updates & #📢┃announcements are about updates. ATM there is no update when the next update is 😂

[amber violet]

Please please please add a "Try again" option to do the same everything (ex: if I was generating part 3, clicked Try Again, so it is like I went to part 2 and re-did with the same words I was just creating for the part 3 clip).

[lofty sonnet]

search by full songs

[placid marlin]

Suggestion if it's not already possible(I didn't find a way). Having a way to shuffle the playlist and/or continuously play songs from the playlist. Currently my playlist plays songs in order and then quits when it gets to the final song.

[raw cave]

Suggestion: Release V3!

[strange barn]

Yeah, that 'last word' thing is pretty rough...good suggestion!

[formal bear]

https://tenor.com/view/release-the-kraken-clash-of-titans-gif-26590372

[raw cave]

Is it happening?!?!

[formal bear]

It is cooking

[raw cave]

https://tenor.com/view/cat-pot-stirrer-pot-stirring-stirring-the-pot-salem-gif-26011679

[delicate cedar]

I don't know why but this one always makes me bellow with laughter.

[raw cave]

https://app.suno.ai/song/bbafce90-e05f-4fa4-bbe9-fcfa11af0ac0

[worldly pagoda]

What are the possibilities to use seed number for generations? Let’s say I found the perfect voice, tone and performance and style .. and I want to use it all the time as my theme song, how can I keep the consistency without clicking on continue every time to create similarities and being cut at the end of the song. A new creation would be better on a seed-based level.

[sage mirage]

Seeds are a much-requested feature

And nobody on the dev team ever comments on it, but ???

[delicate cedar]

Just because something does not get commented on doesn't mean it isn't something considered and thought about.

[hybrid crane]

I talked about it with one of the Devs.. there is a misconception how it works. People think too much as it works in SD.

Suno doesn't offer the option to use seeds because its AI music generation process is designed to be inherently unpredictable and creative, creating unique compositions each time. Seeds are typically used in some AI models to reproduce identical outputs from the same input, which doesn't align with Suno's approach to fostering originality in music creation.

[worldly pagoda]

Then how its logic knows what to use from the original composition when we use “Continue From This Song”

Logically, it’s using something “seed”

[hybrid crane]

Don't shoot the messenger.. I've been told that keeping a seed and having the same sound is a misconspection when it comes to Suno. Continue from this song has to do with memory, not seed

[worldly pagoda]

All the respect 😁🤲🏻 didn’t mean it this way ..

[hybrid crane]

I admit.. I thought the same at first

[worldly pagoda]

Hopefully they will figure it out, because with a seed function! The potentials would be “humongous”

[hybrid crane]

Maybe memory can help with that.. pure speculation here though

[worldly pagoda]

I noticed all the big “text-to-speech” platforms I tried, do not provide the seed function, even though the high demand. Hopefully they will figure it out soon .. the potentials of having consistent AI generations is what we mostly seek and the market demands.

[hybrid crane]

I looked up my conversation and it comes down to this:
The way Suno handles music generation differs from some other AI models that use 'seeds' to create deterministic outputs. In Suno, seeds are not set, which dispels the common misconception that users can control the generation process to produce the same result with the same input. As explained by one of the Devs, even slight changes in the starting point of a song, like starting at 1 second or 0.01 seconds, can significantly alter the resulting generation due to the intrinsic randomness of the system. While there are parameters that influence the generation, the inherent randomness is always a factor, leading to unique and unpredictable outputs each time, even with identical inputs.

[worldly pagoda]

Understood. So if they can afford the “memorized” parameters to be inherited in a new generation, as they are doing it in “Continue From This Song”, and instead of clicking on a function to “continue”, it can start a new generation based on the previous generation, that would be a step forward. Because “Continue From This Song” as much as it’s a workaround, yet would love to have a new generation based on it. Could start a new theme and style with the same performance, then I do voice-to-voice conversion for a voice I already generated/cloned in AI previously.

[gaunt fox]

Is there any way to produce repeatable compositions/vocals - for a brand (mine) it would be very beneficial as a paid subscriber

😦

@worldly pagoda I'm all in on that idea

being able to say use the composition of vocals and beat style from a previous clip would be awesome

[worldly pagoda]

Yes .. this is a game-changer.

[tulip mist]

my suggestion is: Volume slider

[gaunt fox]

@tulip mist also good for quiet sections in a composition = [volume_50]

and then you could add [tempo=120]

but not knowing how this works in the background makes it easy to throw ideas around - the complexity is obviously enormous

[worldly pagoda]

The tempo 🫡

[gaunt fox]

could you use sliding scales too, for intensity in parts - it may be possible? I only started yesterday 😄

[sage mirage]

I figured it was something to how it's built, but seeds, in practice anyway from how I use them and see them used, is to test things

You have a seed so you have something to compare against. I could have a base, then see what terms do

[hybrid crane]

Did you read this too? #💡┃suggestions-web message

[worn atlas]

Maybe a way to simply duplicate some parts should be enough. Continuation with no change.

[sage mirage]

Timeline control and picking and choosing, yeah. Voice keeping would be extremely nice too, like MidJourney does style and now character reference

I'm starting to play more with using a song and builting an entirely new song off the back of it, so even being able to clip the front off a generation would be nice. Otherwise now I have the problem where I have a song with an entire other song on the end with no way to 'pluck' that second song out

All trying to work around the lack of finer control

[worldly pagoda]

How about separating the music/vocal, then bringing the instrumental into Adobe Premier to use “Remix” audio in it. You will have some extension. Or put a musical transition to remove the “cut”, these just workarounds for the moment.

[sleek scroll]

Our own melodies PLease

[thorny coyote]

how do you seperate the music and the vocals becuase i can't seem to find that out

[toxic portal]

That you can play special effects sounds... For example sea waves, a patrol car, a person in the background whispering, a bell or something like that

[turbid yoke]

I would like to ask something. I understand that you are thinking of providing an API in your roadmap. But your terms of use say this:

"In no event will you use the Output to compete with Suno, including to create a competing product or service."

Basically, this is exactly the purpose of an API, in my case, I want to resell music I create or give the option to users to create music along with other services.

[formal bear]

That's kinda the behavoir we want to prevent, right?

Let me give you some positive examples: you are a teacher from a school that want to use SUNO for education in music/foreign language, or you are some organization try to create music to spread good will, and you need to generate that at scale, then yes, an API would be considered possible maybe, and that's not a competing product or service.

[turbid yoke]

So, resell music is not allowed by Suno?

[sage mirage]

Reselling the generation of music is the thing it seems like. You'd be a direct competitor to Suno who's business is selling music generation

You said yourself that you'd like to give that ability. Conflict of interest

[turbid yoke]

Please, I want reply for Suno staff. This is a serious topic.

[turbid yoke]

Is there any concession can Suno make, under your conditions to allow a reseller website?

[sage mirage]

It seems extremely clear to me but sure my guy, knock yourself out.

[turbid yoke]

Sorry Tony, but this doesn't make sense to me. If you are a teacher the 30USD/month plan is enough and if you are an organization trying to spread good will you need a couple of songs. No an API providing songs at scale.
The purpose of an API is precisely become a reseller of your business.
So, that's why, I'm wondering if it is possible to make some agreement since I want to resell songs.

[fiery linden]

That's not the only purpose of API though? There are other uses - for example, implementing it into Unreal Engine so you can generate songs through it. If they specifically say no competing product or service, then it's pretty clear that they don't want you to use it for that purpose. Just like some Stable Diffusion models allow people to use the model to sell image but not for image generation. But anyway, let's wait for a Suno staff to answer for certainty.

[formal bear]

This question is out of my scope -- suggest you email [e-mail] to follow up

[dawn fjord]

Please number pages in the web app forwards instead of backwards. Currently we dont know how many pages we have without clicking through the lot and we cant find a song because its page number is constantly changing as you add more songs.

[sage mirage]

Would be cool to be able to just type a number as well

[turbid yoke]

Yep, Unreal Engine I think is a valid case.
About Stable Diffusion they allow to sell images. So, it can be Ok to sell songs? It's supposed to be Ok since is General commercial terms.
The examples they provide in the terms of Use are more related with bad will or harming people. Anyway, I don't want to be sued 🥺 .

[fiery linden]

Yeah, pretty sure right now you can sell songs you generate. You just can't let other users generate songs through your service using their API. That would be competition to their service.

[open rapids]

BUG

and my Coins????

[placid flume]

We should have a discord room for posting music we love and are inspired by.

[torpid wave]

Here's some improvement ideas:

1. When a full song has been generated, currently it shows the prompt. There could be 2 buttons below "mix" where you could select to see either "prompt" or "lyrics". The creator could use this button to make it easier to follow only thenlyrics, or to see the prompt as we do now. Listeners should be able to report, if there are illegal things in the lyric section, and make the lyric section only visible for payed users to prevent malicious behavior and easier to ban evil people.

2. Playlists: There should be an option to modify the song list, to move songs up and down.

3. Library: a filter where you can choose to see only specific generation; clips, full songs, public. Etc. what comes to mind. Genre? If it's possible.

4. A volume tab. I know, already asked by many I believe.

5.0 Ability to comment songs. 5.1User profile pages that people can browse.

Asking a lot 😄 But I'm also happy as things are now. Suno has improved a LOT!

[brazen root]

It is hoped that there will be API services. In addition, an ultra-short mode is provided, 10-20s, suitable for business scenarios. Do we have a plan for that?

[faint galleon]

It would be cool if we could do some live prompting. alter the music by prompting while it is playing. I visualize future DJ's will be prompting during their live performance creating music on the fly, perfectly reflecting and building up the vibe of the crowd.

[carmine quest]

SUGGESTION : It would be very useful to have the possibility to regenerate any part of a song, with all the current controls, and also to be able to define the desired length, while keeping the coherency with the rest of the song. This way, it could be used to improve any part of a generation that is mostly good but needs some tweaking, and it could also serve to simply cut out a part we don't like.

[sage mirage]

You can rearrange songs in a playlist already. Just click and drag them as you'd like

[pliant canyon]

you can use tools like Ultimate vocal remover or RIPX DAW to seperate vocals / Instrumental ... RIPX DAW can also export midis from Intrumental / vocals of the song so you can reuse this in your DAW

[pastel wadi]

Would it be possible to store a voice seed that when you have found a voice that you like. can be used again and again for new songs?

[hybrid crane]

No.. you cannot.. Because Suno does not use deterministic seeds, you can't save a specific voice setup and apply it consistently across multiple generations. Each song generation is treated as a new, standalone creation. This approach aligns with Suno's emphasis on fostering creativity and exploration in AI-generated music, allowing for a wide variety of voices and styles, but it also means that recreating a specific voice or style exactly can be challenging.

[thorny coyote]

You are godsent

[patent field]

Suggestion: Make it possible to comment on peoples songs, almost like a social media app. I Would love to be able to comment on someone elses work that I think is really good within the suno.ai website. Also the ability to generate one instrument at a time. Say I'm looking for a guitar solo, It would be cool to be able to just type in Melodic Guitar Solo at a specific tempo and key

[sage mirage]

Don't know how far and social they want Suno to go but it might be nice to 'follow' specific users

Some people know how to wield Suno really skillfully

[hybrid crane]

They are working on something like that

[sage mirage]

Rad

Maybe a volume slider too

[hybrid crane]

and V10 while we are at it? 😛

[sage mirage]

Dang I'm cool with a v4 but sure gimmie all the v's

[hybrid crane]

We want it all 🙌

[rugged light]

The idea of seeing Suno as a social media would be fantastic, even more so if there was the possibility of generating revenue with our music and, who knows, getting a discount on our monthly fee 😂 . Regarding the guitar solo, you can try to place the prompt directly in the lyrics, like this [guitar solo]. I just don't know how to change the tone of the guitar 😕

[sage mirage]

What about prompt battles? That's a social-thing that someone like you or the Moderators would run

it kinda happens naturally

The other day when everyone was doing 'What' songs

and doesn't require anything but moderation

[hybrid crane]

Or subject of the day?

[sage mirage]

Yeah stuff like that

I know MidJourney does that

Give people recognition and you'll drown in people submitting

Plus people will just refine their skills since that's just how it goes

[wind abyss]

styles in discord separated channels i like to see. blues,rock,easy listening,🤢 dub,🤢 techno 🙂

[hybrid crane]

#1154790975240155247 ?

[compact forum]

This has probably already been suggested but I'd love to see more ways to edit the music generated in the website. A simple crop tool would be nice since sometimes the ends of song generations kinda freestyles into its own thing and you want to continue a part 2 from the mid point of the initial song. Something else somewhat tied to the first suggestion is the ability to merge your continuations of a song into one whole song instead of 2-3 different parts.

[woeful burrow]

Suggestion: since 1 generation is always 5 credits and each button press always generates 2 songs, I think it would be heaps more user-friendly show e.g. "Generations left: 50 (100 songs)" than current "500 credits". You are showing a number to user that they have to divide by either 5 or 10 to have any meaning at all. That's not user-friendly. I see a lot of other services too have shifted away from showing "credits" and just showing meaningful numbers instead. Credits may be more meaninful to Suno, but not to the end-user.

[fiery linden]

My suggestion earlier is to use an already generated song as reference. Just like how right now you use reference for continuation. If there's a way that you can add reference when creating a new song - and the reference only takes into account of the voice but not the music, then it's possible to create a new song using an existing or at least similar voice. I think it should be doable but we'll see if Suno team wants to take an attempt at this feature. @pastel wadi

[thorny coyote]

When I wanted to add a guitar solo to my song I continue from the place I wanted to put it and changed the genre to ethereal guitar solo and it flowed together. So try that

[fiery linden]

I would like the ability to regenerate but only with minor differences too. I wonder if it's possible to set a % when regenerating. Like Eleven Labs' similarity meter. This would be a nice way to fix small issues like mispronunciation but you like the generated part without redoing all of it.

[potent crane]

for me that would be over-complicating the app, any edits and other adjustments can easily be done in post production in apps such as steinberg's wavelab or other similar apps that are well suited to such actions.

[compact forum]

I'm not sure how that would overcomplicate it. I see it as akin to trimming videos in the camera app on your phone. I just wanna move the slider over to where in the generation I want it to end and it just cuts off the rest.

[sage mirage]

Dear Suno Dev Team:
Negative Prompting would be really nice, please and thank you.

[formal bear]

if we double the credit would you still use it

[sage mirage]

I personally would, yes. I'd probably make the option hard to accidently toggle though, maybe put it behind an 'advanced' option?

I can only imagine the user load having to explain what it does and why it costs twice as much

Maybe even an option you have to toggle on the subscription page

Or heck grant it to power users, idk.

Users you can trust

[hybrid crane]

I would.. I would!!!

😄

[potent crane]

yeah! "no effin' autotune"

[formal bear]

kk good to know

https://tenor.com/view/oh-no-homer-simpsons-hide-disappear-gif-16799752

[sage mirage]

until next time~~~

[potent crane]

Just noticed you can't put a command such as "end" on an instrumental, would be nice to have the choice

[hybrid crane]

Why don't you put it in the music style? I do that all the time.. works great

[sage mirage]

For an instrumental yeah that's where I'd go first, generate a new clip but it's musical style is "end"

or maybe "end" spammed as much as can fit in the box

[inner coral]

Zero to 5 stars system instead of thumb up/down.

[inner coral]

Oh, yeah, I would still use it, probably even more, I sometime use 500 credit just to find the right thing because there is something annoying. I would say 95% of my generation are wasted credit (On V3 Alpha).

[opal pulsar]

Something that would be great is to have a multi track feature to get lyrics and each instrument in different track. That would be so great to create good stereo mastering

[opal pulsar]

But not sure that the AI is trained for that 🤔

[lunar bolt]

It would be amazing if we could upload our own instrumentals and have AI create something that adapts to it. voice

[worn sphinx]

anything planned like have "mongolian throat singing" or "bulgarian women choirs" ?

[acoustic pike]

Ahh.. it takes song/music from SAREGAMA Carvaan. it's not good for me. 😕

[sinful mulch]

I'd like if there were comments available on songs and/or some way to see where the tens of thousands of listens are coming from for some of the songs on the explore page. For example, some way to see the referrer URL that most of the traffic to these songs are coming from since it's entirely driven by direct links. I would like to either be able to discuss the songs on Suno or find the subreddit/forum where people are coming from.

[sage mirage]

Seems more social aspects are coming

[formal bear]

ok your opinions are all heard :D. thx

[formal bear]

thi sis a super interesting one

it's funny some super early mocks were like that, I think 0 to 4 stars or sth

[sage mirage]

I kinda like the simple thumbs up down system, tbh

I'd rather have comments

To articulate why people like or don't like

[formal bear]

youtube? XD

[sage mirage]

But then you have to have moderators

Oh it can go real garbage yeah

But Suno is still small and uh, 'local'

[formal bear]

I think we thought about opening up emojis

[sage mirage]

That could work yeah

[formal bear]

but even that...the amount of 🍆 ...

[sage mirage]

Allow only approved emojis

[formal bear]

yeah then it is like facebook

[sage mirage]

Let someone else figure out what represents dongs

[formal bear]

I like emojis

ok back to work

[sage mirage]

👍

[potent crane]

didn't realise that was an option, I would have imagined that it would try and find a genre of music called "end"

[sinful mulch]

They could just use something like Disqus instead of building their own comments and moderation tools. and they already have moderators on discord, I assume its easier to moderate with something like Disqus because you can just IP ban people, but you can't even do that on discord.

[acoustic pike]

What about Playlist?

[devout sparrow]

I would like to suggest constructive criticism: if possible, adjust so that the songs generated for Brazilian funk are as close to the style as possible. When trying to generate a Brazilian funk song with a specific prompt, I notice that the artificial intelligence often incorrectly associates it with American funk, which results in losing credits until the desired result is achieved. It is important to highlight that in the prompt I clearly specify: "a Brazilian funk song about...".

[acoustic pike]

If we likes some others created song.. we want to Listening again that.. so how can we add it in our playlist?

[acoustic pike]

Indian music?? 😅
Always add 1st saregama carvaan then start the music.

[devout sparrow]

"I would also like to inform you that it is necessary to correct the fact that sometimes the lyrics of the songs report the style of music we are requesting. For example, when I request with the following command: "a trap song about sea waves" , the artificial intelligence ends up covering a trap song about ocean waves in the chorus. This is not desired, and until it is corrected, we will lose a lot of credits."

[devout sparrow]

and in Brazilian music? how do I do?

[acoustic pike]

Kind of Ukelele guitar mini drums music like that?

Youdeling songs

[worldly pagoda]

Hi .. I believe prompts’ weights could be beneficial for the style, example:

Male Voice::2, Female Choir::1 …etc

As I am trying, the results are magnificent, still the output quality could be enhanced to make it useful professionally. (using v3 here - custom)

[olive fiber]

I sent a support email 2 weeks ago.. still no reply 😦

[rugged light]

I write the prompt in portuguese to have Brazilian songs. It has always worked

[finite spade]

As a creator with additional tools at my disposal I'd love to be able to take my creations to the next level. To do that there are a few features I would absolutely love to have available. Here they are:

1. Ability to get a separate vocal track and music track (++ would be able to get individual instrument tracks - but just vocals and music would be enough)
2. Ability to mix instrumental only and custom lyrics - what I mean here is the ability to start a song, then do a continue with this song and add a section in that is instrumental only.z
3. Ability to adjust the overall desired song length - e.g. enter the lyrics, set a time of 3:10 seconds, choose whether to fill time with repeated vocals or instrumentals.
4. Rather than the abrupt end to songs - provide an option to set the time of the last part of the lyrics the user wants to hear before a fade out. I often get something that sounds great - comes together quickly, sounds great - but after a certain point there's repeating of lyrics and then an abrupt cut.
5. Better FAQ on creating custom music. I see all kinds of variations in what people are supplying to the AI in the Explore, but nothing concrete about how to layout your song. Some form of standarization here would be great.
6. A suggestion on the simple formatting of a song might be the following:
   a) Verses - [Verse 1], [Verse 2], etc.
   b) Chorus - [Chorus] and also give the user the ability to do a [Chorus Repeat] between single or multiple verses to reduce the need to re-enter the chorus and potentially end up with it being seen as a verse or a situation where the AI sings it differently.
   c) Bridge - [Bridge] - I often tell a story in my lyrics, bridges are important.
   d) [Instrumental] - I see people using this but have little luck in getting it to work. Would be nice ot be able to put this anywhere in a song or better yet [Instrumental :30], or even [Instrumental Outro]
   Thanks!

[finite spade]

100% on the weighting of style components. Would also love to be able to download the vocal and instrumental tracks separately so I can re-master them / clean them up / adjust levels.

BTW - I should also state - that although I have a lot of suggestions - I want to commend the creators of Suno.ai because just freaking WOW - I've up to 26 pages of songs in my library, loving what you have created. I also realize that for the time that it's been out there the new alpha is LEAPS AND BOUNDS ahead of where V2 was and I gotta say - you are so far beyond so many other generative music AI tools I've used!

[violet delta]

Hello, devs up?

I have a BIG request that is important to me. I am a singer, among other things, but I want to be able to use Suno to create instrumentals that I can sing to. Of course, you can create instrumentals, but those intrumentals are not based around lyrics like regular song creation. So here is my suggestion. When the user wants to create a regular song (with lyrics), give them the choice to generate a version of that song with someone singing, and one without any voices. This way, someone like me (a singer) can use the song with voice as a guide, and then use the backing track (without the singer's voice) to sing to or even mix and master that song with their voice. I'm sure you all have a slew of requests. You all are doing an amazing job. I would appreciate it if you would take this into consideration. I'm a pro user, if that helps.:)

[turbid yoke]

In my case, I sent 2 emails yesterday 😦

[frank anvil]

You write me words. I'm a singer, too ...

[hybrid crane]

I understand your need as a singer to have instrumental versions of songs. However, Suno currently creates music as a single, mixed audio track, which includes vocals, instruments, and other elements all combined. Technically, separating the vocals from this mix to create a standalone instrumental track is quite challenging. It's not as simple as just removing a layer; it involves intricate processing to isolate specific elements from a completed mix, a feature Suno doesn't currently have. Your suggestion is insightful, though, and points towards a significant area for potential development.

@frank anvil ^^

@finite spade my answer above partially answers 1 and 2 for you too

[frank anvil]

Thanks for your reaction and telling about technical background.
But I'd immediately' upgrade to Premier Plan, if it would be possible the download separate vocal and instrumental tracks. Perhaps in suno.ai V4? Never losing hope ...
Although tools like Ultimate Vocal Remover exist, they often yield subpar results, especially with noise/trashy sounds like in Garage Rock, which I frequently generate. For example, With suno.ai I created a song I really love and wanted to record my own vocals, but the instrumental quality suffers after vocal removal, particularly where vocals were extracted.
What I need, and I believe many other musicians too, is the ability to further process and customize generated songs. THAT would be a Premier Plan!
So let's hope for the future!

[hybrid crane]

I am not saying to lose hope. I am just trying to be transparent. Trust me when I say that the Devs wants to deliver this too. It just takes time to figure out ways to do that. The "will" is there.. the big question is "how", since this is new tech 🙂

[frank anvil]

I know - I'm an ai teacher, doing courses, understanding a bit of the background. And now that I know you are aware of the need to get separations, I can tell this to the clients asking me. Thanks!

[spiral acorn]

Is it possible to make the title of the song affect the track image? Nearly all my tracks have variations of that bird you get when you do an instrumental based song 😆

[hybrid crane]

it's not logical for most people tbh... Maybe they should give the option though

[spiral acorn]

doesn't it nearly affect every single instrumental?

or is the bird coming from my styles???

[hybrid crane]

I personally would think we should get an editor to refresh the images and lyrics.

[spiral acorn]

are the styles affecting it too?

[hybrid crane]

Yes 😄

[spiral acorn]

ahh, that explains it

well that's not as bad as i thought then

[chilly current]

PlaylistsThank you for adding this very helpful feature!

The following are just requests.
If there is room for further consideration, please consider them.

It would be very easy to use if the Playlists to which each song is assigned could be displayed within the frame of each song in the list of Clips.

[nova hatch]

a little suggestion for THUMBNAILS :
i noticed that all my finished songs and most other ones have what i call the OUTRO BIRDIE because suno always picks the last image and i usually use [outro] in the last take. how about picking the 1st one, or even more nifty let the user choose which one of all the ones used in a session ? every suno list would look way more funky and all the songs would have a more personalized touch.

[hollow locust]

Have 'Create' and 'Library' both go in the same chronological direction. It's very disorienting to switch from one to the other. Better yet, give the user the option if they want top-to-bottom, or bottom-to-top chronology.

[bright fjord]

It would be great to add more Carribean music genre ( KONPA, ZOUK, SOCA, MERENGUE).

[midnight herald]

I'd love a more accurate and realistic electric guitar and guitar techniques later down the line. (Palm mutes, alternate picking, Tremolo picking, legato, string skipping, open chords, resonating notes, pinch harmonics, octaves)

[raw cave]

Just put it in the style of music....
https://app.suno.ai/song/d9996df3-6b15-4ab3-8817-fc1abe8a8dc7

[bitter yoke]

The V3 is amazing.

But a feature I know people doing this professionally would 100% pay extra $$ premium:
Add a model with higher quality and sampling rate, Its perfectly ok that it takes more time to generate each song, I would be willing to wait even 1 minute if that makes it feasible.

[opal sun]

Being able to tag tracks and search those tags would be awesome for folks who are generating hundreds or thousands of tracks.

Turning the bugs and suggestions channels here into forum style posts would help curate ideas better and reduce duplicate bugs and suggestion posts. Not sure if this server has those options yet though. ❤️

[raw cave]

Here's a thought...in my a.i. imaging program, I can take two pictures and "paint in" the joining/merging of the images. Would be neat to take two clips of different genres and be able to join or merge the clips together.

[bright fjord]

But konpa is not yet

[bright fjord]

For Konpa i got country music....

[raw cave]

I do rap and get rock. I do polka and get bluegrass. You have to keep rolling the dice.

[raw cave]

With style of music: Konpa
https://app.suno.ai/song/42784d1d-7841-4aba-8e6b-dd301db882a9

[raw cave]

With style of music: Konpa. brass, slower rhythms of méringue de salon, pulsating beats of tanbou
Sounds pretty good if you ignore the lyrical hallucinations!
https://app.suno.ai/song/067735e5-71a8-4671-98a5-a1d020ad58a2

[sage mirage]

It would be really nice to be able to tell Suno to look further back in the song chain to pay attention to the, say, first musical part of a song and use that as a template for a generation later. Songs I work on will have a cool musical bit at the start that gets lost and Suno can't ever seen to 'grab' it again

Not even editing related or anything, just a sort of focus

I don't know how far the temporal aspect of Suno works when creating new clips

Get 3-4 generations into a song and you just sometimes lose the musical style. The base and beats usually stick around

[leaden marlin]

what i hate about suno and continuing songs is that sometimes i will get the perfect generation for like the middle of my song but because there is silence at the end it wont continue and then all the generations i get that don't have silence aren't as good

[hybrid crane]

Are you on the free plan? Because with V3 you can continue from anywhere. So then you can continue before the silence.

[shadow orbit]

Is there a way to go back and see all of the songs you have liked from the explore page? In the Library tab you can filter down to the liked songs that you created, but I want to see all songs I liked that others created. Thanks!

[hybrid crane]

Not yet.. Great suggestion though 🙂

[sly seal]

Wish i could cut in the end of a song since the AI often ends the song in a good way but it is annoying when it 10 sec later starts a new song.

[raw cave]

What blows my mind is when you have a clip (1) with verse & chorus, then an instrumental or solo clip (2) and then return to a verse & chorus clip (3). If I alter the chorus in clip 3 sometimes it still repeats the chorus from clip 1. How does it do that if I'm continuing from clip 2 without any chorus to reference?

[sage mirage]

I'm messing with this right now in that I have a song starting with a normal, non-musical conversation between two people

and at least with this song, I was able to summon that conversation into the mix

I copied the same lyrics and structure and all that, along with the same tags that I used to make the spoken word part

and Suno went "hell yeah" and mixed it in

Same voices and even cadence

This song I'm working on is the most I've ever felt like I'm standing over a bubbling cauldron

Sure I'm throwing in eye of newt but ???

[weak elk]

I appreciate the big "THIS SONG WAS MOVED TO TRASH" text, but maybe not right on top of the player?

[tropic roost]

Is there an affiliate link I can share?

[sage mirage]

There is no affiliate links at this time

[sage mirage]

Another suggestion, probably for the future?
I've started to relax and let Suno just kinda do what it wants, with the lyrics and hallucinations people always complain about, and I think I actually can appreciate how desperate it is to sing and fix things, you know? Whenever it sings it sings right and correct, seems like when people get mad at it it's because the machine decided otherwise.
IF you change that behavior later I would appreciate it if I could still let Suno take the reigns more like current v3 Alpha does. Lots of people want more direct control and they most certainly have that. It's just this is a new sort of creative expression and working with it in unique ways is cool and I'd hate to see this go away.
I guess a sort of 'strength' setting or something similar that lets Suno have more of a say.

[rugged light]

I think I've already suggested this, but here we go:

→ Maybe, in the future, earning rewards, like a discount on the subscription with your own music inside Suno, in a kind of 'make your music, earn points for each listener and like, and exchange for rewards'. I believe this would take Suno to the next level.

→ Who knows, a contest with the best generated songs. The 3 winners could receive discounts according to their plan. For example, people on the free plan could have the chance to get a discount on the first monthly or yearly subscription. While Premier or Pro users would also get discounts + extra credits. It would be a way to attract more subscribers and maintain an interactive environment 😊 .

[bright fjord]

yep a bit

[raw cave]

Just gotta finesse it

[bright fjord]

You rocks, because i'm haitian descendant, and i really would like to make some.

[bright fjord]

But it's very difficult to make suno make one properly

[raw cave]

Sometimes, like with the second prompt, you have to help the a.i. a bit and describe the music/instruments/emotions/tempo that you're looking for.

[bright fjord]

yep i think so.

[hybrid crane]

As far as I understood yesterday from a Dev, if we want more control like a negative prompt, you will have to keep in mind it would probably cost double in credits.

[bright fjord]

But i already waste a lot of credit on lyrics that has been skipped etc. So for something that Suno that doesn't recognize i won't too long for the moment

[raw cave]

Welcome to Suno! Join the club!!

[bright fjord]

I can't wait for the final version of V3, when all those problems wil be ended

[raw cave]

Would be cool if the devs gave like an overview of how Suno works...without exposing proprietary coding, etc. Would like to get a high-level understanding of how the prompts work and how the a.i. ultimately generates the music. For instance, the double credits for a negative prompt...does that mean the generation is created and then goes through a second generation to remove/replace elements?

[hybrid crane]

If V2 perfect? I don't want to dissapoint you, I am a poster girl for Suno, but even ChatGPT still has it's problems. AI is a beast that we still have to learn to love and tame. V3 will be beter than Alpha but we will still have things we have no control over.

[bright fjord]

I means the problem with skipping text. Like for exemple, it keeps skipping my text, i decided to move the text where it jumps....But it jumps far away again 😂 .

[hybrid crane]

I understand your curiosity. I have it too. But I understand why they are not telling us too much. Suno is ahead of the game, and if you give a glance behind the curtain it is for us "hocus pocus", but for other LLM engineers it will not. So I doubt they will anytime soon.

[bright fjord]

Praise be upon GOD, it finally work after around fifteen tries😭 😇

[hybrid crane]

It's not easy to find the right balance for the engineers. So I don't envy them to be honest.

[hybrid crane]

I praise the engineers...

[bright fjord]

Yeah i know, we got to encourage them.

God creates the engineers that creates Suno that ours prompt generate musics🤣

[hybrid crane]

Let's not bring theology in the scientific mix 😉

[worn atlas]

At least, know why when i put Konpa in the style i am getting à bossa nova 😉 . Would like devs to give a near exhaustive list of styles specifically exotic ones : Brasilian, african ...

[raw cave]

There is no "list" of genre/subgenre that Suno does/doesn't make. Everything is a toddler a.i.'s interpretation of the type of music you request and is never exact.
Did you see my attempts at Konpa?
#💡┃suggestions-web message
#💡┃suggestions-web message

[safe wadi]

It will be good to be able to keep the melody of a song we generated and be able to

• Start from the beginning
• Change the style of the music
• Reset the lyrics
• Remove some instrument to the existing
• Add some instruments or voice to the existing at a specific place
• Keep everything the same and allow us to modify a specific element at a specific part (Like the lyrics)
• Be able to associate different part generated into one track by specifying where we want to make the join (Lot of time different generated part suggest amazing things but we can't use them for different reason. so to be able to put them into the main track by chaining part that we like will be good)

[bright fjord]

https://www.suno.wiki/faq/style-and-lyrics/styles-and-genres/

[bright fjord]

https://www.suno.wiki/faq/style-and-lyrics/styles-and-genres/

[bright fjord]

It will be in V10

[raw cave]

Correct! "Crafting" a song is a good way to put it.

[vocal thicket]

its wild to me how many people complain about the price when $30 is dirt cheap for what its actually doing

i would gladly pay double for negative prompts

[feral aspen]

I would gladly pay 4x for doubled positive and negative prompts. (Seriously, it should work and just take 4x as long...

[vocal thicket]

yeah the price point is not an issue at all for what you're getting

[devout sparrow]

I propose a review of the pricing policy for music generation. I suggest keeping the cost of 5 credits per new song, while halving the amount for generating parts of a song. I note that I currently spend considerably more credits on parts than when creating a new song from scratch, which results in excessive expenses. Therefore, it would be more advantageous to charge 2.5 credits for song parts instead of the current 5 credits. The current billing system is discouraging as it often results in expenses of between 50 and 100 credits per song

[vocal thicket]

now if they can find a way to make it so the voice doesnt keep sampling itself as the song gets longer, leading to more of the AI-esque autotune being thrown in, lord have mercy

[frank anvil]

That's absolutely right! I also would pay the double - for getting stems! Some day the will learn how to do it ...

[vocal thicket]

and you get what 10k credits for 30 dollars? you're complaining about 100 credits for a song? 30 cents?

brother

the amount of resources its using to generate that song

[devout sparrow]

Don't just focus on your currency, my friend. I deal with the Brazilian real, where 30 cents for you can mean almost 5 reais for me. Let's be generous; This opportunity is for all of us.

[vocal thicket]

im focusing on the cost that it takes to power what's going on the backend

[hybrid crane]

Remember, Suno provides a unique AI-powered music generation platform that requires significant resources for maintenance and development. Their prices are extremely fair.

[vocal thicket]

i would not be surprised at all if its break even / losing money and the price point is just so they can get a bigger dataset to train ti

[devout sparrow]

Are you making money from this? It looks like he was offended.

[vocal thicket]

im offended because your entitlement is annoying

i have nothing to do with suno whatsoever, im a game developer and I hate when people act like you do

look at it from the other perspective, its not all about you and trying to get it as cheap as possible my man

[hybrid crane]

Starts ups often are not making money... or even invest to get a costumer base.

[devout sparrow]

If you are not part of the team, I don't understand why you are involved in this matter. I didn't invite you to this discussion, and besides, it's important to learn how to respond to a customer. A customer usually does not have technical knowledge and it is essential to treat them with respect. Take SUNO support as an example, who demonstrated politeness in dealing with me and clarifying the facts. I understand now, not because of the lack of politeness, but because of the courtesy of the SUNO attendant. Avoid getting involved in questions you are not competent to answer and, most importantly, avoid interjecting where you have not been asked to.

[vocal thicket]

stop crying about pricing then, you're annoying and unhelpful

[devout sparrow]

Thank you for your response and education, friend.

[devout sparrow]

Be mature, focus on developing your character, and treat others politely, especially those who do not have the same knowledge as you.

[hybrid crane]

Okay.. can we all stop this destructive way of communication?

Both sides are heard, and we never make peace and understanding if we start throwing mud.

[worn atlas]

I suggest doubling the Price for a longer song generation. 10 credits for 3mn . We will have better consistency. The continuation is a credits consumer.

[sinful mulch]

I'd like a way to mass download my songs whenever I ctrl+click or shift+click to mass select them, or a way to download an entire playlist of my songs with one button.

[devout sparrow]

The intention was to get this point across without causing offence. However, instead of contributing positively to the community, as this boy just did, the user only aggravated the situation unnecessarily, as he allowed his immaturity to offend him.

[languid finch]

hey! Don't know if thats been suggested already.. it would be great to also be able to download stems, like drums, vocals, melody, bassline seperatly. would that be possible to program? 🙂

and a MIDI file as well

[hybrid crane]

Yes many many many agree with you. But not easy to achieve. Suno is investigating this.

[languid finch]

oh cool to know that they are looking into this! 🙂 thanks!

[worn atlas]

I was wondering if a reliable list could be provided by the devs : What is known as "always working". Searching for what should work is not free.
A very large list of styles/genres to be tested :
https://www.musicgenreslist.com/

[sage mirage]

Nothing is always working is the thing

[hybrid crane]

Just like for us, it is new for them too. They train the models, but don't have all the answers yet.

[worn atlas]

Not true. If you select "rock" or "reggae: you will always obtain that. And this is the "official" suno documentation. The issue is for more exotic styles. "World" styles

[sage mirage]

There are times where even rock won't go that direction is the thing

Depending on how far you are into a song

What prompts you are using, other tags, how your lyrics are structured

[worn atlas]

I know. I would like they update a "reliable" list as they go. The frustration is when you find an awesome music and you are not able to reproduce it as you dont know how it is labelled by suno

[hybrid crane]

The team is still small and there are a million things to do. But I know they are working on getting thing more tested. I just have to say that KakerMix is a bit right. There are no certainties. Not yet at least. Hench ChatGTP still lies now and then..

[worn atlas]

For chatGpt, prompt engineering has evolved a lot to better control it and if you use the API seting "temperature" to a low value will lower its "creativity" ... and you have a constant cost if you dont use the API for unlimited use.

[hybrid crane]

Chat GTP started when? 2 years ago? Suno started with music just a half year ago

Time will solve everything though 🙂

[raw cave]

Broad genres (rock & reggae) are easy to match because 1) there is so much music to reference and 2) so many songs will fit into those genres that your preference tends not to be as specific. When you get down to subgenres the sample size of music gets much smaller and your preference tends to get more specific.

[worn atlas]

I know. Suno is really awesome . I have said several times here that the tool is a technological disruptor. So we are somehow frustrated to not have what is possible. But it is just a question of time and patience.

[raw cave]

I'm sure many people may have put "dance" in the music style and got Konpa, Samba, Cumbia or Bachata and had no idea what they got.

[worn atlas]

Yes but i am surprised of the high quality produced by suno. But it is impossible to reproduce what i got because i dont know how to recall the style.

[raw cave]

Nothing here is reproduceable...every creation is a snowflake.

[tardy bloom]

One thing I wish Suno did was when you continue a part of the song, the style of music and most importantly the voclas stay the same. The number of times i got an amzing first part but then the second part changes everyhting slightly which isn't bad but then the thid part also changes and by the end theres a buf differnece between the first and last part. Does anyone else have this problem or is it me that's doing something wrong? I'm still new to creating music with Suno

[vocal thicket]

its because its sampling the vocals to reproduce it and slightly adds more correction to them every time, the vocals will always get worse the longer the song goes

thats my only gripe and its a bitch to fix, but if they make it possible where they split the instrumentals from the vocals it should resolve that as well

[hybrid crane]

I personally do not recognize this. But the only think I can think of is that your verse are not consistent in sybils? Structure is everything 🙂
(unless you mean distortion, which happens sadly)

[worn atlas]

Again i dont agree. When i speak about reproduceable of course i am only speaking about the style. I am able each time to get a fine reggae root style. It works for me 100%

[tardy bloom]

@hybrid crane That makes sense I'll try to make the syllables for each line the same as they were in the previous part. Also is there a way to cut out the last second or millisecond of a song out?

[hybrid crane]

Are you a paid sub? Do you mean "continue from anywhere"?

[tardy bloom]

No unfortunately I'm on the free subscription

[hybrid crane]

Aw.... well with the paid version, you can continue from anywhere, and that would solve your problem I think?

Then you can continue just before it goes wrong, starts to sing gibberish, when it is fading out etc..

[tardy bloom]

Ok thanks, I'll try the syllables thing as well to see if that helps

[brittle geyser]

could you make an update separating instrumentals from vocals, and instrumentals into individual tracks? Midi too?

[delicate spire]

I have the paid version and I wish I could be able to seperate vocals from the instruments

If i select generate songs with lyrics I get better instruments but if i only select instruments it generates some goofy stuff in my opinion

[bright fjord]

https://www.bandlab.com/splitter

[delicate spire]

I tried that and it sounds so terrible after I do that

cus its already low quality

[bright fjord]

ah...

[olive fiber]

They should have a tier for paid support

[gaunt wind]

I would like it if the song didnt cut off before the 2 minute mark if it didnt need to. many songs of mine stop too early and even halfway through a sentence... https://app.suno.ai/song/bf8d753d-c9c2-49bf-a2f3-d82f010afa4e/ for example. There should be tags that the AI explicitly follows like [intro][break][outro][end] and it only ignores them if it runs out of time.

[tiny sluice]

I got good result with https://ultimatevocalremover.com/

[rotund orbit]

You know something I don't need but would love to have? A take tree. Sort of a way to backtrack what takes or partial takes went into any given song, and almost be able to backtrack and redux the evolution of any piece. It would be really nice. And probably not that hard to code. Unless it's written in Java, because everything in Java is harder than it needs to be.

[karmic snow]

Ability to use any "voice models" with suno, basically using any voice you want.

[devout sparrow]

Would it be possible to create an option to add an intro to the song lasting up to 30 seconds? For example, the song is generated and then I have the option to add an intro of no more than 30 seconds before it. It's important to point out that the 30 seconds are just an example, and the implementation will be done to the best of our ability by Suno

[gaunt wind]

someone else mentioned making a song instrumental to start, then continuing at 30 seconds with lyrics and vocals selected. havent tried it but could work for now. but yes more flexibility with generation (before, and instead of sections, and between) would be great.

[sharp arch]

**i think it would be VERY good idea if there was like any option or something where it reminds the voice from a song that u made and choose and then it makes the new song u want with the same voice so u can make more music with one voice **

for me its there

[hybrid crane]

I see how you're feeling unheard, and it's crucial to clarify why Suno currently doesn't support user-uploaded models. Suno, especially in its V3 Alpha stage, is still developing its core functionalities. Integrating user-uploaded models poses significant challenges in terms of compatibility, quality control, and resource allocation. It's a complex feature that the platform is not yet equipped to handle. Suno's team values user feedback but must balance it with realistic developmental capabilities and resource limitations. The AI music generation field is still in its infancy, and Suno aims to be transparent about what users can expect at this stage. If V3 Alpha doesn't meet your needs, V2 offers a more stable alternative. Remember, Suno encourages continuous feedback but also urges users to maintain realistic expectations and explore current features optimally.

And if you are missing option in your menu in discord, it is due to the fact that you closed them.

@brittle geyser @delicate spire I understand the demand for a feature to separate vocals and instrumentals into individual tracks, commonly known as stems. This functionality is highly sought after for greater control and creativity in music production. However, providing stems is a complex process involving advanced AI music generation technology, which Suno is currently not equipped to handle. The team is aware of this popular request and is exploring future possibilities, but as of now, it remains a challenging feature to implement due to technical and resource constraints.

Suno is currently scaling up their support team due to high volumes of inquiries. If you have any specific questions, feel free to DM me here, and I'll do my best to assist you.

[hybrid crane]

I understand the appeal of having a specific option to create intros for songs. While Suno doesn't currently offer a dedicated feature for this, you can creatively use the 'Continue From This Clip' feature to first generate an intro part and then continue with the main part of the song. This method allows some flexibility in designing the song structure, although it might require a bit of experimentation to get the desired result.

[hybrid crane]

When songs are ending before reaching their full potential, one strategy is to increase the amount of content in your prompts. More lyrics give the AI more material to work with. If you find that your lyrics aren't sufficient to fill the desired duration, consider adding elements like [instrumental] to extend the song. Instrumental sections can help fill any gaps and contribute to a more complete musical piece.

Furthermore, exploring different music styles can provide valuable insights. Different styles can influence the AI's interpretation and generation process, potentially affecting how it utilizes the provided content and the overall song length. By experimenting with various styles, you might identify certain approaches that lead to better outcomes, giving you a clearer understanding of how to guide the AI for your desired results.

[hybrid crane]

The concept of a 'take tree' for tracking the evolution of songs is currently being explored by Suno's development team. They're looking into ways to make the creation process more transparent and navigable for users.

[rotund orbit]

That would be so cool

[hybrid crane]

The idea of using any voice model you want in Suno is intriguing. However, Suno's AI works with a set of predefined voice models and doesn't support importing custom voices. This limitation is due to the complexities of training AI models with specific voices, as well as ethical considerations, like respecting voice rights and avoiding misuse. Suno aims to offer a diverse range of voices within its capabilities, but custom voice importation isn't feasible at this stage.

I personally have 2 premium accounts.. thus I am paying double and have double points. That's a way to do it 😉

[alpine niche]

hello, any suggestion or tips to make just speaking voice without music

[hybrid crane]

For creating just a speaking voice without music, Suno initially had a feature called Bark, specifically for speech generation. You can still use Bark for speech synthesis on the platform: Hugging Face.

[alpine niche]

thx

what about with music in suno？

[hybrid crane]

What do you mean?

[alpine niche]

like making an intro part sounds like film conversation

[hybrid crane]

With spoken word? You can try things like [spoken word] Infront of lyrics that needs to be spoken. And also add things in "style of music", like podcast, or radio announcer. I've seen people successfully use that. But doesn't always gives the expected results though.

[alpine niche]

yeah, i have seen someone do that either, just forget how they achieve that, thanks!

[hybrid crane]

Here an example from a user: https://app.suno.ai/song/6f2e4614-8e0b-4240-a52a-921bd9d19665

[alpine niche]

great! thanks!

[hybrid crane]

He used this in his music style: poetry reading, passionate, intense, articulate, Black American male speech, spoken-word poetry

[alpine niche]

I found out! but thanks for the illustration. Another question, is bark possible to clone voice and make it to rap?

[hybrid crane]

@feral aspen is the expert in that case... Could you give some pinpoints here Jonathan?

[feral aspen]

It's difficult to do both at the same time, clone a voice (that isn't rapping) and then make it rap. And just getting it to rap is hard. I think Suno music is better, just prompt "spoken word" and related terms

[alpine niche]

thank you for answering my question, could you tell me how to clone a voice with bark? I did not find the entry on that website.

[feral aspen]

Just cloning a voice is not an official Bark function, but there are community developed models that can do it partially, so you have to look for bark forks with the feature

I should point out, you do NOT need to use voice cloning to make Bark voices, to make good Bark voices, or even to make good Bark voices of a particular voice style.

You only need to use voice cloning to make copy of real world voice, of exact audio file

[alpine niche]

thank you！

[spiral acorn]

~~Suggestion:

• Ability to archive songs
• That way you can clear your workspace up a bit without having to delete generations~~

I've noticed when doing continuations I will try a lot before I get the one I like. But at the same time I don't want to delete the ones I don't like. If there was some kind of "send to archive" that could be quite useful

Edit: Ignore this, I realise I can use the "thumbs down" for this!

[safe wadi]

Suggestion:

• Add a Metatag [Restart] To keep same melody but reset the song and the lyrics. Restart will put a break (no sound) and just after intro > verse > Chorus
  Like that in a way we will be able to correct bad lyrics and keep the same melody as before

[weak elk]

v3a is sometimes REALLY has a hard time following the lyrics

[pliant canyon]

well it could be possible somehow, I had some generations wich only created suddenly "dry" vocals by accident, sure, this are unusable .. but it could happen ...shown in this example at the timeframe 0:12 it gives a short impression of the current quality what to expect on dry vocals... https://app.suno.ai/song/b88131ec-2276-4413-ad3e-dab0bbf2ad63

[hybrid crane]

You can also do that semi with adding "a cappella" into the music genre. But that still is not the same as splitting into stems 😉

[pliant canyon]

Ahh okay.. 🙂 the splitting into stems after a song is mastered is not helpful anyway... better would the seperation before the mixing process starts... for all included instruments or at least the instrument groups (bass, drums, other (pads keys), vocals) for clear stems... but without any insight over the generation process this is a pure speculation

[hidden vault]

PLEASE add the habitability to add a intro into a clip generated that starts in the middle off verse. I have (and many others) potentiality GOD songs but cant generate and intro for them and they start "mid-song"

[rotund orbit]

Request: It would save me a step if the audio was sampled at 44 instead of 48. Also wav exports would be nice too

[timber halo]

@hybrid crane do you know if Suno use some sort of AI A&R to feature songs in the Explore section of the website? If not, that is my suggestion.

[weak elk]

Suggestion: the image generated for the song is obviously modeled after the lyrics. But if there are no lyrics, there's always a bird. It could also use the Title for more Funs. 🙂

[tepid trench]

I'd like some image variety for instrumentals, and also some kind of auto title if I don't put one in (I usually don't).

[pliant canyon]

suggestion: project folders with songname for all generated parts for a song

[sage mirage]

You can get mostly there using playlists in a similar way

[spiral acorn]

yea atm the only way around is to not use the lyrics section or include phrases like "(moon in sky)" in the lyrics section to generate a Moon image. Often it won't vocalise these phrases if you have "instrumental" in the "Style of music" section - though sometimes it still will

[rain sage]

Looking for some constructive criticism on my first 4 songs. I own the music rights, and these are my own original song lyrics. Thanks for looking.

[spiral acorn]

post this in #💬┃general-chat or #🐣┃suno-showcase

[rain sage]

https://youtube.com/playlist?list=PL998NRrqhR_Un8V9lo5JvURNO_B90c_vO&si=_AeYBdxvXscVoXfO

[spiral acorn]

you might get more feedback in #💬┃general-chat

[rain sage]

Ok thanksn

[foggy light]

Suggestion: API with commercial use of songs generated via it

[dusty dove]

My suggestion to the SUNO project:

• Load a saved song made with SUNO and from there you can continue or make a remix.
  I think it would be a good idea.
  What do you think?

[unkempt folio]

First can I say that is freaking awesome! I was able to take a song idea and your program brought it to life in a way that I never thought could be possible! Just WOW!! What I would love to see is finer tuning of a segment. For example, focusing on part of the verse of the song because its not quite right, or changing the tempo of the song. Also the ability to harness or create a band from that song. For example I really love one of my songs and now I want to create a template or band image so now when I load that template it's going to write and sing song with that particular style and sound.

[lofty sonnet]

Remix or Continue From This Song functions when you select the song

[unkempt folio]

Also a project folder to keep your songs separate.

[lofty sonnet]

Can be tedious but, playlists in library feature exists now

[lofty sonnet]

Can continue from this song, then use prompting to have the song make an abrupt switch or change into a new idea

[torpid jetty]

Welcome! For help, check out Suno Help

[lofty sonnet]

not perfect though, love that idea

[unkempt folio]

Yeah, it would be nice to have project folders. Since we are creating songs. We could create a band profile. Under that band profile project folder you create your songs.

I mean we are literally composing music with the help of AI. Talk about leveling the playing field.

[lofty sonnet]

style presets for sure

[unkempt folio]

Yes, I create a song where I love the style and sound. I want to capture that and basicaly make it into a whole ban.

Eventually adding AI avatars that will sing it

[lofty sonnet]

keep playing with continue from

[unkempt folio]

Oh that's a great idea!

[lofty sonnet]

change prompt, pick an early part or transition point

[unkempt folio]

Yeah I will defintely play around with it.

[lofty sonnet]

I would like to see custom style presets, especially negative prompt presets like dont ever give me pop generations, etc.. Seems I sometimes get "Pop" when I'm not asking for it..

[unkempt folio]

What style are you doing?

[finite spade]

@hybrid crane Thanks for the info on that. I was able to use another AI application (lalal.ai) to extract the vocals from the instrumental and played around a bit with it's capability to extra each of the instruments as stems. It was pretty impressive and did a great job for what I needed. Suno definitely is the best for generative music I've experienced thus far though. I had dinner with a notable musician last night and showed him some of what I was creating and he was very impressed.

[hybrid crane]

That's great! It's true there are several "splitters" around. And Suno could easily implement something like that, but what most people here really want is it to be done pre-processed and not post-processed, to avoid the noticeable bleed. Most users (and me too!) would love to have studio quality. And I have a lot of faith in Suno... It might take some time, but the Devs are bright people.. they will figure it out 😉

[tepid trench]

A splitter that can also fill would be nice. By that, I mean where you remove vocals and/or drums and it leaves noticeable gaps. It'd be nice if there was a way to patch things up to sound better.

[spiral acorn]

Would it ever be made so the credit usage is based on the generated length of the output (rounded up to the nearest credit)?

Maybe it's done the way it is for a reason (like the AI still uses the same power regardless)

[sage mirage]

It's because the credit system is built around having a GPU at all, not how much you use the GPU

Trying to do it based on actual usage has no real chance of working in the market

It's use it or lose it

[spiral acorn]

i mean the generated output length

like for example I generated a 30 second track

(which is what I needed actually, so it was the right length)

[sage mirage]

That's just it

[spiral acorn]

like 2 minute track = 5 credits
so 1 minute would be 3 credits rounded up

[sage mirage]

Whatever the system is built for

The length it generates is a model thing, I'm almost certain of

and then that has to be run on GPUs that Suno rents from whoever

[spiral acorn]

yeah i do get that part

its more like a nicety

[sage mirage]

Would be nice yeah

[spiral acorn]

what made me say this is i'm generating a lot of continuations and in this case a lot of them are like 40-50 secs (which is what I want)

[sage mirage]

I've just taken that as how Suno works

You aren't going to be 'efficient'

Credits are really attempts

And since I am trying to make a complete song, some attempts 'carry' more than others

[spiral acorn]

that is a fair point

[sage mirage]

Sometimes the 10 second generation that I pay for is worth 5 2 minute ones I don't use

[spiral acorn]

what I will say it is VERY cheap lol

[spiral acorn]

yea I have had that kind of thought

[sage mirage]

It is very cheap for the amount of serotonin my brain gets

Which is why when tony asked if negative prompting cost twice as much if we'd use it

and a bunch of us fell over each other to say 'yes'

'cause it would cost 'more' but it might not actually cost more

[spiral acorn]

yea, though that is something you can opt into right?

[sage mirage]

If we can get what we want sooner

[spiral acorn]

ahhh

[sage mirage]

It would have to be yeah

[spiral acorn]

yea gotcha

[sage mirage]

I suspect it's because you are asking the system to work harder since you'd be double-prompting

What you want, and what you don't want

Specifically my use-case, right now, is generating the conversations at the start that I've been playing with, Suno desperately wants to sing'

The more practical thing I suspect people would use it for would be to steer Suno out of the vanilla pop zone when they make their yodelcore transwave witchouse

[karmic mirage]

Sorry to butt in, but is it possible, afte you have created and completed a song to create another song using the same artist/voice on another song?

[sage mirage]

Not currently, no.

[karmic mirage]

okay, thank you so much.

[zenith atlas]

I use the Web Paid version of Suno.... ON the EXPLORE TAB, would be good to see if you can search for songs By View counts??? Like see most Popular Song etc if that makes any sense..

[raw cave]

https://app.suno.ai/song/dd9b1d40-db77-4ac1-b15e-28f868fc7602

[agile coral]

So, getting a surprise rap verse in a song has been someting of a white whale for me, and I think I've hit the limit of how close I can get. I did some experiments, and what they show is that suno is very strict about what type of vocals are allowed in what genre.

https://app.suno.ai/song/0c555dac-576a-4374-8a5d-4ae688d14073

https://app.suno.ai/song/68008263-b6f7-41da-97cd-e3d0e672317f

In "coffee shop love," I use a continue to keep the backing track while telling Suno that the style is changing to hip-hop. After that, I use a continue that only hears rap lyrics, but the association between vocal type and genre is so strong that it switches instantly. Even though "rap" is in the prompt.

In "You Will Obey Me, Machine" I achieve the same thing, but suno changes to EDM-style vocals halfway through the stanza even though the style is still supposed to be hip-hop!

I've done a bunch more experimenting, and all of it is consistent with my theory here. In places I can hear the conflict, such as a monotone section and the "character" sounding like a white boy rapper. If suno is this strict about other genre conventions, it explains why getting genre fusions requires so much coaxing.

Hope these findings are useful in improving suno!

[zenith atlas]

thankyou ill have a look at this

[agile coral]

Okay one last experiment worth showing: The behavior is so extreme that even "synth" in the prompt is enough to make suno refuse to rap most of the time. It has to be specifically hip hop.

https://app.suno.ai/song/01428f07-927e-4f1f-8599-e745b3c69bce

[sage mirage]

Not sure the back end work to make this happen, but would it be possible to snip the ends off songs? Clips? Just pick when it ends? Similar to 'Continue From' but instead it just ends there rather than generating a new clip

That way you could end a song whenever you wanted, and even if Suno generated more unrelated stuff on the end it wouldn't matter since you could cut the clip and just end it instead

[frank anvil]

Yeah. Something like a AI sound repair function to fill the gaps (as you called it) with something like "assumed sound".

[polar pivot]

I have problems with several songs on V3, if for example, I like the first verse of the lyrics but I don't like the rest, I continue the song on the time code before the part that I don't like (and change the lyrics). But then comes a problem, Suno keeps ALL the text in memory and starts to hallucinate more and more as we advance in the song.. Could we access and edit the lyrics of the previous generation when we continue a song? (Also, I don't know if the system is seed based, or if there is an equivalent to "cfg scale", but it would be nice to have more control in general)

[novel idol]

iOS app?

[haughty valve]

Ability to see listens even after song has left the explore page

[hexed wharf]

please add search function. thank you

[gaunt wind]

Ok so a bit weird, but it would be cool if we could use our mic to sing/hum/whistle a melody and then suno works out the notes and timing and adds that to our songs?

by which i mean not our actual recording, but the melody, the gist of what we did

[still marten]

Volume Slider pleeeease 🙏 🙏 🙏
It's extremely inconvenient to go to the system mixer every time I need to change loudness for the next song.

Suno is becoming my favorite music streaming platform
So volume control is a must
Just like it's hard to imagine Spotify without ability to change volume

[hollow locust]

For the 'Autoplay' button to stay off permanently.

[lyric temple]

Adding styles to choose from. It would be great to use when exploring and creating songs.

[hollow locust]

Option to download videos in 16:9

[rustic burrow]

Hey loving Suno 🙂

Was thinking it would be very helpful if the "continue from this song" outputs were nested under the original track. So instead of it being in the default list of everything, you would be able to collapse / open the sub-tracks which were continued from the original

[rustic burrow]

Also as another UX suggestion, when I delete a generated section, the red popup covers the center of the UI and blocks the ability to play songs for a moment. Would be a minor change but helpful efficiency improvement. It could pop up on the bottom right, or top right

[last sundial]

Hi all! I wonder if I can supply my voice to train the model (am a singer on Spotify with 30m+ streams and 250k monthlies) and who I should contact about this?

I'm blind! Found the email 🙂 Cheers

[hybrid crane]

Congratulations on your success as a singer! Regarding your question about training Suno’s model with your voice, it's important to note that Suno currently does not have a feature for users to contribute their own voice models. This is partly due to the complexity of voice training in AI and potential legal and copyright concerns.

[last sundial]

Got it! I'll keep an eye out and email the team just in case 🙂

[pulsar coral]

Devs!! Can’t download files from mobile iOS please help and fix!!! Best…

Files are only direct link

There’s no package contents

[tepid trench]

I just mentioned this in #🪲┃bugs-web - it's like "download" is actually "share".

[pulsar coral]

Will try

[stone monolith]

It would be fantastic if one could download music in uncompressed formats such as WAV and AIFF, instead of the very cheap 192 kbps MP3.

[pulsar coral]

It won’t allow you to download/convert from web ui to phone. Could be an Apple iOS. I’m sure Android devices work. I work for AT&T and we’ve applications that don’t run or work on iOS versus Android is more user friendly. It doesnt unpack the file from the web link.

[tepid trench]

Android also doesn't download, it wants to share with another app (see my screenshot in #🪲┃bugs-web )

[pulsar coral]

I tried saving into notes and text messages

But it keeps it as a link

Is there an app i can save into?

[tepid trench]

Copy the link into a web browser

[pulsar coral]

Cool, trying Google drive as well

[tepid trench]

Excuse the fast MS-Paint effort here 😆 . At the risk of possibly overcluttering the mobile menu, this is what I think would make sense.

I've renamed "Sharing" to "Visibility" since that seems clearer and the trash option is further down.

It might just be better to have more icons next to the song itself though (like/dislike/share menu/download menu/trash and then "...")

[pulsar coral]

@tepid trench so what you were mentioning didn’t work what I did was download Google Chrome then I switched to request desktop site. Apparently it does not work on mobile at all neither android or iOS but if you change it to desktop site via Google Chrome extension on your phone or tablet you can then download.

It’s working now.

Google fixed it

[wind lintel]

Can you make with voices from Portuguese from Portugal? ( It's a different Portuguese from Brasil)

[hidden vault]

Idea 💡: You could add to every generation info like BPM or key. You could use free software like https://tunebat.com/Analyzer. It´s been a great thing for me to learn about music and have more control off what i want. It´s probably not difficult to do and would be awesome!

[hybrid crane]

That's not easy. I feel your pain because I have it too with Dutch. My neighbor country speaks Dutch too but we a different account. Same goes for Chinese dialects. Suno Devs are aware but have no solution atm (yet).

[hidden vault]

Another idea💡 : in Pc when you delete a song, the message to undo is over the music player, making you wait to moove or change songs. Moove the undo message up and thats solved!

[weak elk]

Suggestion: search box for your library

[pseudo surge]

Please give us a folder/sorting system for all songs. Is this going to happen eventually? I'm tired of having over 30 pages of parts of songs that I can not tell which go with which other parts. Most of course are before renaming came in. But either way I want to be able to move songs around from folders to other folders or something similar for better sorting. I'm sure everyone who is making a lot of songs would appreciate being able to sort and order their music. Especially those that are neat freaks like myself. 🥺

[pseudo surge]

Thank you for posting that link and also yes I agree with this. There are times we want to make songs better or work with to make sound better but is difficult without this info. How accurate is that site you linked? It gives the info so fast.

[sage mirage]

You can currently use playlists to do this

It isn't perfect

But it's far better than looking through pages of generations

[pseudo surge]

I would have to go through and add every single one to a playlist and it doesn't actually move it away from the creation page. I don't know, it is not the same as actually being able to sort and create order. I'm trying to go through and clean out the parts and songs I don't want and it would be nice if I could take each one and move them somewhere else out of the way.

[sage mirage]

I am not making excuses for Suno, to be clear, just trying to offer solutions you can use now.
I have many, many pages of generations going back to v2 that I will never use again, I just treat those as lost things deep in those pages you are talking about.
I will like clips I like in order to give me a quick marker for stuff worthy of saving while I'm working. I'll name clips that are especially nice, and if I find a clip I like to start a song with I then make a playlist just for that song, and dump every clip I make in service of that song to that playlist. I'll finish the song, then move the finished song into a new playlist reserved for complete songs. I usually then clear out all the clips, maybe even delete the playlist entirely.

[hidden vault]

in my experience is pretty acurate at least with bpm that is the thing i search for with this 🙂

------------------------------------------------------------------------------------------------------------- 💡 IDEAS to add in the WEB💡 ----------------------------------------------------------------------------------------------------------------------

1. In Pc when you delete a song, the message to undo is over the music player, making you wait to move or change songs. Move the undo message up and that's solved!
2. The ability to add intros to clips that start directly in verse. Shouldn´t be impossible cause you already have the ability to continue a song using the stile of other clip and you can stick 2 clips together (to make the full songs) keeping style and making flawless transitions. You only have to figure how to do it backwards, but it would be the same as creating a continuation clip calling for a complete silence [silence/mute] and then call for an [intro] and the rest of the song, stick all together and then cut everything before the silence.
3. Saying this: the ability to trim cut clips, and to try different clips sticked together like a puzzle making the AI do the transition between them would be CRAZY good. So a simple editor would be awesome.
4. You could add to every generation info like BPM or key. You could use or partner with free software like https://tunebat.com/Analyzer. It´s been a great thing for me to learn about music and have more control off what i want. It´s probably not difficult to do and would be awesome!

(respond this with more ideas for the developers and i´ll add them!, let´s make Suno better together🤠 )

[pseudo surge]

So after you complete a song and download you usually delete it from Suno with all its parts?

[steep jackal]

All V3 of 20 seconds should be refunded.

Seriously.

[sage mirage]

I never fully delete anything

But I let it get 'lost', I guess

Or save the parts I really REALLY like to its own playlist

[steep jackal]

V3 won't even read lyrics anymore

[sage mirage]

But I can always clip what I want from the main, finished song

[pseudo surge]

Yeah that is useful. I just started trying to do this in FL Studio but you first have to have an idea of the bpm before it will tell you and I guess it is acurate but now I have a second source to check from

[sage mirage]

Since even the spereate clips are going to be stuck with a start I can't get out from under, you know?

I don't see a reason to keep my library clean in Suno. I just organize the top part I am working in, I have the songs I like saved in playlists to go get them again

It's a different way of doing stuff for sure, but alas

@hybrid crane

ol' spammer

Oh they are everywhere you certainly see em

[hybrid crane]

@feral aspen

Thnx @sage mirage

[wide belfry]

Something I'd like to see in subsequent versions is the ability to do Negative Prompts.
-Hum, -Distortion

[tepid trench]

Some previous discussion regarding the suggestion of negative prompts (including the possibility of it costing additional credits) --> #💡┃suggestions-web message

[potent crane]

Get's requested almost every day, it HAS to be introduced, maybe like other apps, only to premium members (as an incentive)

[potent crane]

I agree, been using V3 constantly, and one thing it is less good at than V2 is lyric recognition, it misses whole lines, misses words, repeats lines, repeats whole verses

[delicate cedar]

Why though?

The devs have been remarkably upfront in admitting that it would make no difference in the actual output since any distortions are a result of the way the AI generates the audio and not in any way related to the compression of the music format. MP3, WAV, as far as we've been told it makes no difference. It will sound the same.

[potent crane]

But it will be uncompressed if it's wav - are you telling me there's no difference in quality between mp3 and wav, if so then it has been the biggest con of the last 20 years

[delicate cedar]

No that's not what I said.

I said with the music made by Suno there is no quality difference between the MP3 audio and the WAV audio, according to the devs because the music is created in whole, any perceived distortions or imperfections in the "compresed sound" are there from the inception of the file and not actually due to compression, according to what they have said on this subject in the past.

[kindred carbon]

Hmmm, I'm inclined to disagree ^. It might 'sound' the same, however there is a lot more DATA that gets created when exporting a .wav. This in turn gives producers the ability to do destructive editing before that data is lost and therefore we can do a lot more with it at a higher polyphony. Right now I upsample the .mp3 to fake the data and to give the audio more to work with when I do destructive editing (separation/isolation, effects, vocal clean up/master, etc).

It's not necessarily about the final 'sound', as you're right the human ear can't tell the difference. However if I play you what Suno spits out vs one of the treatments I give it once I package around it with my own samples/drums, it's a HUGE difference, and it all stems from that initial data loss that is then compensated for.

[hexed minnow]

UX/UI: Scrolling the user to the bottom of the Create page when a song is done is very disorienting and jarring. Instead of interrupting the user's workflow to notify when a song is done, use a status message "<name of song> completed". This keeps the user in the same orientation.

[sage mirage]

There is no extra data, though. Suno outputs directly the quality you get. If it did it in .wav there would just be a whole lot of extra file size for no gains.
This isn't like normal music production, Suno gives you the final master.

Normally the way it works is how you expect. Suno doesn't work like that.

[delicate cedar]

Exactly. The artifacts, the noise, the hisses, whatever it is you're hoping to avoid with uncompressed wav files doesn't work because those artifacts, noise, etc are a part of the original audio and not compression.

It's like thinking you can record an old record and by recording it digitally avoid the scratches on the tracks. It's a part of the track itself at this point.

[kindred carbon]

That's unfortunate. I suppose I don't understand how th algo works, it's just. . . . . if you guys brought it this far, it's a shame this wasn't though about initially but that's just my bias for what I'm using it for. You'll never buy a soundpack for producing that are .mp3's, so this is coming at it from that POV. If the hiss and artifacts are just a time-stretch for better training material to be used to avoid it, and or better/thicker drums, then cool. If this is full throttle limits, I think a '.wav stem creator' with higher bit rate all around including clean sources - would be a cool next gap to fill.

[potent crane]

If there's no difference, then why does a similar app AIVA offer wav's as a premium perk, whilst the peasants have to settle for mp3?

and are you suggesting that the dataset used by suno is comprised of mp3's? Surely the source material are wavs?

[delicate cedar]

Without commenting on what other companies or systems do, it seems to me that if the developers here say that the quality is in the model output and not the compression, I'd be inclined to take their word for it.

You don't think they could do an easy slam dunk by releasing a special credit cost to get everything as WAV? Or creating a new Music Producer tier and escalate the cost but provide WAV files to that tier only? I'm sure that enough people would jump on it to make it financially worthwhile...for a month at least.

But the developers feel that to do so would be a disservice to their users. Kind of nice to have that kind of honesty and candor.

[pine tree]

out take
rating
five star
Right now, all we have is a good or bad rating.

[potent crane]

I hope you are right, and let me be clear, I am producing a heck of a lot of songs here that I am very happy with, quality wise.

[kindred carbon]

I'd happily pay to try the .wav's, but be just as fast discredit it if there is literally no difference or it's just the mp3 upsampled for a profit slam dunk :). I take your word for it, and think in time this gap will be filled here or elsewhere but you can't blame users for pushing for it or wanting deeper understanding as to why. AIVA sounds horrible if I was to compare and I used it for a few yrs, I believe they just upsample the mp3 anyway.

[kindred carbon]

Devils advocate: But this implies you guys only recorded old dirty tracks as your source/training material lol? What I CAN do, is record it at 96k/32 bit float, and this gives me a lot of DATA to do destructive editing and yes improve the overall master/track/eq/cleanUp harshness hiss clicks and scratches (RX/WAVES/Lot's of tools do this).

[delicate cedar]

I am not technical with regard to how the models were created, simply repeating what has been told to me on a general layperson level.

[kindred carbon]

I totally appreciate the back and forth and the understanding I'm gaining - so don't see this as me challenging anything I'm just after the best for you guys :). EDIT: And for what it's worth, you guys are bleeding edge as it is, so thank you for having created such a thing - I'm already seeing V10 and where this might go/branch off to.

[delicate cedar]

The truth is that I have privately advocated for WAV release as a kind of an upgrade and a way to differentiate the tiers of the plans available. Well, I guess not so privately any more...

But if what I have been told is correct, it would only be a placebo and not in any way ethical given what the devs have explained is how things work. I suppose with less scruples it would be easy to just do it any way and make people happy without actually providing value.

I'm kind of happy that we have ethical developers though, aren't you?

[sage mirage]

The jump from v1 to v2 to v3 has been massive, and I suspect it's more to do with the models themselves and not by some sort of file format

I suspect v4 will be the same sort of leap

[kindred carbon]

Of course, if you guys ever need a guinea pig to do actual testing (GIVE ME THE .WAV LET ME DO MY THING WITH IT TO TEST IT PLEASSSSE), I'd be willing to pitch in my .02 whether it's placebo or not. Dev's vs producers tools/perspectives/datadigging may be different to real world application and testing.

[sage mirage]

Someone else said that when v2 came out, normal people went "wow this sounds pretty good" and music people went "eeehhh"
Now v3 is where the music people are now going "wow this sounds pretty good"

I'm coming at this as a plebian user though that just knows what sounds I like

Not someone from production

[rose seal]

Will keep you in mind! From our testing, WAVs are not noticeably better audio quality and still retain the graininess (otherwise we would put them in a paid tier).

[kindred carbon]

Thank you, and looking at that screenshot above, a higher bitrate mp3/wav may not sound a ton better but if it's possible it does give people a beefier file/data to manipulate so not from a sonic pov but an editing/post pov it would be very beneficial (most producers are going to be separating these into full layers/arrangements vs using Suno as a full editor/tying final versions together. I also believe YOU guys are going to birth a lot of producers as a side-effect of having to edit these files).

[sage mirage]

That is true. I've messed with slicing and cutting before but now I'm going even further because the thing I get from Suno could be better with just a little tweak. Which then turns into a larger thing and on and on.

[formal bear]

ok I know this request for high quality wavs, but I also agree with the C-suit decision that wavs are not a step up and we don't want to give our users "fake" good stuff.

As a super secret backdoor, if you take a concat (only concat) mp4 and separate out that audio track, that audio should be made from a wav file directly.

[kindred carbon]

Here's a quick look at a canvas once I am done with it. I've named all the layers so you can see what SUNO spit out and then was separated into, then my own layers added on top of it as to up the overall quality and mastering up to .wav for commercial distro.

I believe everyone here should get their hands on a free DAW (MAC has Garage Band installed free, Audacity now has layers/free). As you can very easily save credits and use the daw/separation tools to tie your pieces together, use instrumentals from one for all, fix tempo changes, move words or full lines around, fix syllables, etc. etc.

Just thought this might help both end users, and the DEV's as to why a higher bit-rate is important when it comes to destructive editing/post production when your source is a .mp3 (which I then upsample to fake data at up to 96k 32bit and get to edit easier without further loss/degradation). Now you can actually improve the sound a LOT as you have much more dynamic range and editing without artifacting.

EDIT: Not only that, but now I can add fx/EQ/master what suno gave me much easier, even get rid of clicks, hiss, mud, harshness, etc.

EDIT2: After using Suno for a bit now, my BIGGEST wishlist ask is to produce acappellas so I don't have to separate and have clean vocals to package around.

Finally - for anyone curious, I have now gone FULL CYCLE. Produce on Suno > Post produce in DAW > Master for DSP delivery > Turn on content ID for my tracks > Successful delivery to DSP's and YouTube with content ID on and all is well re copyright/flagging/etc. (using my own lyrics in Suno that is).

[ornate mulch]

If you do any work with EDM or electronica in general. I suggest you check out the Genopatch option of the VST Synplant 2. Essentially you feed it a sample and it reverse engineers it into a synth patch. Very useful in conjunction with Suno

[mild spear]

why it doesn t give a search option to search songs from others ?

[hybrid crane]

It will in time, still much to be done. It's a relative young company, and a small team.

[mild spear]

Alright thx but for a small Company respect and its a nice work

[formal bear]

yeah it is on the todo list...I even know who I want on the team to work on search, only if we can convince him to join...X.x

[sage mirage]

how about a fruit basket

Are you looking for a particular web-dev sorta person? It isn't me but I have a friend that might be super interested, though I don't know how far you want to go with looking outside wherever it is you want to look

[formal bear]

nah I am even just looking for that person. he is so good that i just want to work with him...

[sage mirage]

If I did want to pass your info on would it just be [e-mail]?

[tepid trench]

A friend mentioned Synplant 2 to me recently. There are some YouTube videos on the Genopatch feature. It seems well-suited to short/simple sounds but struggles with more complex sounds. I might give it a try anyway. Seems like it might be really useful to take a generated part, split it into stems and then use Genopatch to reconstruct the percussion, bass, and possibly lead instrument.

[rare bear]

Add a function to select specific songs and stitch it together to make a whole song out of

[wary hazel]

proszę o wprowadzenie zmian; sugestia w piosence:
[on]
[ona]
[chór]
żeby można było po końcu linijki wpisać daną sugestię, kto w danym momencie będzie śpiewać.

please make changes; song suggestion:
[he]
[She]
[choir]
so that at the end of the line you can write a suggestion about who will be singing at a given moment.

or in what language

( [she - english] )

example:
I'm not here! [She]
You're not! [He]

[formal bear]

yep

[leaden marlin]

Suppose my subscription has expired, would I still retain the right to use the music I created during the subscription period for commercial purposes? Or does the expiration of the subscription affect the commercial use rights of the music generated during the subscription period?

[delicate cedar]

https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448#47fc7b6ad1114a1597903b11c73ade77

[devout echo]

Love the default is now 2 verses and a chorus. I don't like how the two generations now both have the same lyrics.

[delicate cedar]

If you made it while having a paid subscription you retain the rights to songs that you made during that time.

At least that's how the FAQ reads to me.

[ripe mango]

Suno Studio. For iPhone and Android with integrated Studio features

Ability to change voices with Suno presets [Suno library of voices only] so we can use [Male/Female] in lyrics 🥺

Kinda of how what Bark has.

Ability to Adjust Instrumental on the music Scale For example [D-Minor] or [e3-Piano Key] , [G3-Guitar]

[leaden marlin]

Does that mean the rights remain with me even after my subscription period ends. thank you

[sage mirage]

Yes.

[spark verge]

Kinda bummed it generates 2 songs with identical lyrics now, I much preferred when it would generate 2 completely different songs. Is there any way to make it do that again? Or if not, at least have the option?

I was gonna pay for credits again but seeing that made me not want to bother :/

[feral aspen]

Sorry, this was a recent change. But you are not the only person who preferred it the old way, have been quite a few others suggesting the same, so maybe could change in future... (I havew no insight on what Suno will or won't do, but they do pay attention to popular feedback.)

[spark verge]

AHH good to hear that other people feel the same way. Cheers!

[rare bear]

Add a feature to extract the instrumental and the vocals from a generated song

[worldly falcon]

Probably already in the works but a way to have your account linked in your music without needing to change the song's name to include it. So that you can get recognized when listening to the song and people can hear the songs you made public

[potent heron]

Stems stems stems please

[worldly falcon]

An app for mobile so we can listen to music without the need to use a web browser

[civic kayak]

it would be great if you offered for the premium users to download the separate song stems in WAV files and the vocals as dry(without fx) . I'm a music producer and this will help a lot.

[tepid trench]

WAV apparently would not make a difference in terms of quality. There was some discussion about this a few hours ago: #💡┃suggestions-web message

The vocals and instruments are generated together as a single audio clip, so there technically aren't separate stems it works from. This would also include any effects that are present.

Would be nice to have a stem splitter though, rather than needing external tools.

[marble fox]

a question chat to ask questions, for example about the right person

[candid salmon]

API

[hybrid crane]

on roadmap

[compact hollow]

I wish we could chat with the file like ChatGPT to make changes

I wish our credits got refunded if we downvoted and have not downloaded the song

[compact hollow]

Wish it did jazz Lofi and more nuanced genres better

[dim hearth]

this is unlikely to ever happen. generating a song takes a lot of computing power, which has to be paid by someone.

[compact hollow]

Just a wish 😁

[hybrid crane]

I can come up with better wishes then 😏

[lofty whale]

could the V3 come to copilot soon? For free?

[dim hearth]

copilot is microsoft. so unless microsoft buys up suno, no

[lofty whale]

SUNO is in copilot

V2 though.

[dim hearth]

huh... i stand corrected. maybe later then when v3 isn't premium exclusive

[compact hollow]

🙄

[lofty whale]

suno via copilot is currently around 30 seconds maximum if I'm correct, and also limited to V2. But I really like the V3 electroswing music. And would like to make 1 song myself to test it.

[static dawn]

Anyone here into Dark pop?

[pseudo surge]

Well I think it would be cool if we had like small amounts of credits like lets say .5 or 1 come from songs we dislike (but only if we don't download) that then goes into a pool or savings for us to withdraw when we reach 5 or more credits. That way we get refunded credits back for bad generations but not so much that Suno loses much money out of it.

[pseudo surge]

I don't want Suno to become so mainstream that everyone is using it. 🥺 I feel special being able to use V3 and come out with sick songs lol If everyone is doing this then producers will become obsolete, even if some of us are Suno producers. 😅

[hybrid crane]

https://www.rollingstone.com/ TOO LATE

😉

[hybrid crane]

You are special.. you were here before it will explode.. you know more, have experience..
But this is too good not to blow up 😄

[hybrid crane]

This is the suggestions channel.. come and mingle in the #💬┃general-chat channel to find people who are into that 😉

[frank anvil]

THANK YOU! That's something I was waiting for. Really!

[hybrid crane]

the thing is, it is created by the community. So officially not written by Suno. I guess that is why...

But... I know Suno is working on a whole new faq that goes deeper

[worn atlas]

Give comprehensive error messages .
Today suno will give system error when something went wrong in the lyrics. When à style does not exists or a weird combination exists in the style prompt suno will not give a message but will generate something that was not asked by the user. The user expects a comprehensive error message in all these cases. This will help us not burning too much crédits first and secondly will enhance the user's experience.

[indigo island]

hello

[reef bronze]

Add paypal payment

[acoustic pike]

Please Add Solo Song (as Male or Female Singers Voice), Duet Song (for Both Singers Voice) and Background Singers As Chorus Singers Voice Option in Music Style.
Or add
'M' for Male,
'F' for Female,
'Both' or 'together' for Male and Female,
'Chorus Voice' for Chorus Singers
in Lyrics writing option.
Please add this. Thank you.

[balmy pollen]

It would be nice to be able to continue instrumental music

[woven cradle]

The ability to continue instrumentals already exists, at least for subscribers.

[balmy pollen]

I’m a subscriber, maybe a limitation in v3 alpha

[woven cradle]

You can continue an instrumental in v3 alpha.

[shell cloud]

Can the FAQ page contain links to all the songs on the homepage?

[tepid trench]

I don't see how that would be useful

[tawdry zephyr]

PLEASE ADD A VOLUME SLIDER FOR THE LOVE OF GOD

[balmy pollen]

For some reasons, when I try, the continue button is disabled…

[tepid trench]

Type something into the "style of music" box

[balmy pollen]

I tried all of this already, if it work for you to continue a 2 min instrumental song than it must be a bug….

[quartz copper]

That only happens to me when I run out of credits. 🙂

[desert whale]

Hi gang, loving the new playlist option and increase in song length. Would be absolutely amazing if there was an option to remove the vocals after a song has been generated. Continuing on from a song and changing it to an instrumental is great but is limited in that it often fills the space where the vocals were with some form of lead instrument - which once again is awesome, but it would be even more awesome if there was an option to turn "whole songs" into instrumentals of said whole song whereby the only the vocal is removed and not replaced with anything.

[inner dawn]

Hi @formal bear congratulations on the Rollingstons Mag feature! that's pretty epic! We used Suno to showcase the advancements in AI Music at an event we had. We're wondering, will there be an API we could use to bridge Suno with our offerings (use suno through our website or a client's website)?

[inner dawn]

Hi @feral aspen hope your day's going well. I was wondering maybe you're able to answer the question above please?

[feral aspen]

The only word from devs so far on API access is that in the plans/roadmap eventually, but not available yet. You can email [e-mail] and ask about any more specific questions or deals, but on the API generally that's all we know so far.

[formal bear]

on the roadmap -- not available yet

[waxen anchor]

Hello! Most fun to use tool since midjourney! Thank you so much for the hard work.
I really like how suno is able to come up with melodic vocal lines, it would be really convenient, to be able to turn off vocals after generation, so you could take a refrence of how vocal melody is structured, and then perform it yourself. I wonder if it's planned (or even possible).

[tepid trench]

The vocals are generated within the music itself so at the moment at least the only way to do this is to use an external website/program to stem-split, though this isn't perfect. You can also try to use the "continue" feature and try to adjust the style/lyrics to see if it will generate another part with the same chord sequence etc. without the vocals.

[thin tide]

Can anyone tell me if I want to change the lyrics that has been spawnd, how do I make it happend?

[swift jetty]

Bring a feature that allow you to record your own voice and use that in the songs. and also the ability to add music u made yourself to train the bot.! for example if i have 10 papiamentu language songs ill be able to add them to the bot and help the bot understand papiamentu language better?
also sometimes a beat is great and the voices are great but what they are saying is bad and u want to use same voice same instrumental again but cant.
remix should allow u to use same instru and same voice. yet different flow of singing.
or add a different remix call it redo with same instru or something i dont know.
i have many songs that have great instrumental and voice but there is mistake in them so cant really use them.

[steady scarab]

May I suggest a few enhancements to the user interface for an improved experience?

1. It would be beneficial if, upon loading a finished full song into the player, a hint indicator could navigate between the recording sessions (parts). This feature would clarify which part the player head is currently at during song playback. It helps users easily identify specific parts of the song used, even if they choose not to modify them.
2. Another useful addition would be the inclusion of a color indicator choice beside the name/title to differentiate between different productions of songs. This feature enables users to track a specific song through all related parts generated, especially after multiple recording sessions.

I have a couple of other suggestions to enhance long-session experiences, but these are the most significant. I understand that the platform is still in beta, and these features may already be planned or in progress. However, I wanted to share my thoughts.
Additionally, I've observed that when I input a part from a song generated with only an instrumental and then attempt to continue with lyrics attached, I often need to toggle the instrumental button on and off twice to enable the lyrics effectively in the output. It's unclear whether this is a browser issue or something else, but addressing it would be appreciated.

[topaz dagger]

Develop a VST plug-in similar to what ACE studio AI are doing:

Not necessarily voice-centric but having a VST plug-in where one can create a midi and then tell the AI to turn it into vocals/instrument/beat

That would be neat

[dusky fog]

would love it if there were more controls (stem separation, both before and after generation), more tools to help production (bpm, key etc).

Basically, make it easier to go from this: https://app.suno.ai/song/26145070-2b77-40e8-8abc-c3186107e28f

To this (i loaded into Ableton, added drums, tweaked, added some effects, etc): https://on.soundcloud.com/TLoGbYdoRrqbdYEv7

[sage mirage]

Dear devs
Great job on the Explore tab, no notes.

[true hinge]

add more payment methods like paypal. I dont use credit cards.

[dusty dove]

What about 100 or 150 FREE credits per day? 50 is not enough ever. Thanks.

[hybrid crane]

quote of one of the Devss

[dusty dove]

What is this?

[hybrid crane]

you are on a free acount I assume? You get 50 credits per day. The 5 credits are from yesterday when 1 generation failed and gave you back 5 credits.

[dusty dove]

I intend to sign up for the paid plan on the SUNO platform as soon as possible. The budget is already in place.
I'm still in the testing phase, and still improving some ideas to see if I can fit SUNO into my ideas.
Soon the SUNO platform will also have major improvements and that's good for all of us, I believe it's time to add SUNO to my projects.
For now, I'm just a curious user of the platform who believes in its potential.

[exotic jay]

hello, the subscription plan in not updated to the discord account. How do i fix this? credits shows free plan and not the paid one thus, no credits to create. please fix this problem and let know how to fix this asap!!!!

[hybrid crane]

I think is helpful in your case: #📚┃suno-school message

[dusky fog]

Would be amazing to have a command like the midjourney "describe" command. upload a track, get a style prompt out.

[delicate cedar]

It really would. 😤

In the meantime, so long as the artist and the song is known to the AI you can somewhat approximate the band's themeatic elements and the song's style by careful interrogation of one of the ChatGPTs that exist. You can also get hints with Sonoteller.ai which will take a YouTube clip and analyze it for the themes, BPM, singing vocal range, the types of instruments, etc. It's pretty powerful.

By using this combination and some careful prompting you can get pretty close to the sound you like.

[dusky fog]

right right...cool. just figure every model is different so would be interesting to learn what this model is all about! I'll try your suggestions, but tell the product people/person to put a +1 next to that feature request!

[plush owl]

Português

[delicate cedar]

https://discord.com/channels/1069381916492562582/1154768599827890296

[hidden vault]

----------------------------------------------------------------- 💡 IDEAS to add in the WEB💡 ---------------------------------------------------------------------------

1. In Pc when you delete a song, the message to undo is over the music player, making you wait to move or change songs. Move the undo message up and that's solved!
2. The ability to add intros to clips that start directly in verse. Shouldn´t be impossible cause you already have the ability to continue a song using the stile of other clip and you can stick 2 clips together (to make the full songs) keeping style and making flawless transitions. You only have to figure how to do it backwards, but it would be the same as creating a continuation clip calling for a complete silence [silence/mute] and then call for an [intro] and the rest of the song, stick all together and then cut everything before the silence.
3. Saying this: the ability to trim cut clips, and to try different clips sticked together like a puzzle making the AI do the transition between them would be CRAZY good. So a simple editor would be awesome.
4. You could add to every generation info like BPM or key. You could use or partner with free software like https://tunebat.com/Analyzer. It´s been a great thing for me to learn about music and have more control off what i want. It´s probably not difficult to do and would be awesome!

(respond this with more ideas for the developers and i´ll add them!, let´s make Suno better together🤠 )

[steep jackal]

As soon as V3 finds out what you want -- it takes consistent steps to do the opposite.

[worldly falcon]

A way to permanently delete connected links as well. Right now if you permanently delete bad clips or tests, if you or someone else has the links they can still access the deleted songs

[dusty skiff]

Hello.
It would be great if you move the label on the screenshot to some place without interactive UI elements 🙂

[cinder berry]

I suspect this is a tall order, but, it would be nice if continuation could be performed on non-suno uploaded snippets of audio

[tepid trench]

You can guarantee there'd be people trying this with copyrighted material so it'd be necessary to detect this somehow.

[frank anvil]

I made a GPT in ChatGPT having this functionality - I just write band & song, and it generates the style in short terms.

[worn atlas]

Which one?

[vestal sinew]

be great if it had more ethnic or indigenous styles. i tried getting V3 to do a digeridoo and over and over it kept giving me xylaphones or pianos.

[potent crane]

"add to playlist" should be replaced by two commands "move to playlist" or "copy to playlist"

[vestal sinew]

yeah adding intros would be great and being able to mix genres a bit better

[potent crane]

V3's biggest flaw is the lack of intros, and the ingnoring of commands such as "guitar solo", I would say I get a proper intro about 5% of the time, and a solo about 15-20% of the time, it should be 90% plus in both instances

I have lost potentially great songs because of no intro, sometimes I can bolt an intro on in post from an instrumental passage later in the song, but it's a gamble

[vestal sinew]

yeah songs do tend to drop straight into the lyrics and this often breaks the rhythm too as it tries to cram words in at the start.. after a while it starts to even out but it would be good if it understood some formal commands that make it break up verses and stuff a bit better. i realize that its a bit of a random dice roll and the dice roll does help create a good "loop" to make people spend more credits which the cynic in me makes me wonder if this intentionally built into their business model.. i hope that isnt the case but far too many tech companies take inspiration from gambling pokie machines and do things like that to keep people using them.

[sage mirage]

It would be nice if the 'Continue From' option could play the clip from that position. That way I could more easily fine-tune when I wanted it to cut if I could control it from the 'Continue From' area