#📚┃suno-school
1 messages · Page 52 of 1
should look like this
thats what i imput into suno
ignore the male prompt lol it half works
im trying to find the absolute best prompt for it . i know some geners are femail dominate
so putting it be for instance "male swing-step-step-step" will give you 90% chance while removing it will only work 30% of the time
alr i got smth that should be okay
dam
that was an credit consuming endevor
giver heck
its more tuning the "whole " sound of the music
not just guitar
hence the hz
it does matter
i need some explanations on if the system even works in the first place
check this out, this is what I asked for
high catchy melodic complex speedcore track, based on XI (for rhythm game purposes)
this is what chatgpt gave me
[note: Song details and metadata follow]
[note: tempo: min: 220, max: 240]
[note: bar length: 4/4]
[note: tonic: Gmin]
[note: tuning: E A D G B e 440hz]
[note: genres: Speedcore, Hardcore, Melodic Hardcore]
[note: styles: Catchy, Complex, Energetic]
[note: moods: Intense, Thrilling, Uplifting]
[note: arrangements: High-Speed, Complex Rhythms, Catchy Melodies]
[note: instrumentation: Heavy Synths, Distorted Bass, Fast Drum Machines, Energetic Leads]
[note: production: High-Quality Recording, Bright and Punchy, Complex Layering]
[note: structure: Intro, Verse, Pre-Chorus, Chorus, Drop, Breakdown, Verse, Chorus, Bridge, Drop, Outro]
[note: dynamics: Intense and Fast-Paced, Catchy Hooks, Complex Rhythms]
[note: emotions: Excitement, Rush, Euphoria]
[note: chords: Gmin, D#maj, Bb, F]
[note: Vocals: Optional, if used, should be high-energy and distinctive]
[note: Vocal prompts: energetic and intense, catchy and engaging, fast-paced and rhythmic]
this is what it gave me
more commas
tuning commas between the chords and hrz
change bar leanth 6/8
or i dunno try 1/1 lol
see what it does
alr lets suppose i do those things
cuz like
i pasted this
and it generated some kind of
texas cowboy song or wtv idk
couldn't be further away from what i'm asking
so i'm really confused
nope, this time it generated a romantic song lmao
so heres my question why is this not working at all
try this [note: Song details and metadata follow]
[note: tempo: min: 220, max: 240]
[note: bar length: 4/4]
[note: tonic: G minor]
[note: tuning: E A D G B e, 440Hz]
[note: genres: Speedcore, Hardcore, Melodic Hardcore]
[note: styles: Catchy, Complex, Energetic]
[note: moods: Intense, Thrilling, Uplifting]
[note: arrangements: High-Speed, Complex Rhythms, Catchy Melodies]
[note: instrumentation: Heavy Synths, Distorted Bass, Fast Drum Machines, Energetic Leads]
[note: production: High-Quality Recording, Bright and Punchy, Complex Layering]
[note: structure: Intro, Verse, Pre-Chorus, Chorus, Drop, Breakdown, Verse, Chorus, Bridge, Drop, Outro]
[note: dynamics: Intense and Fast-Paced, Catchy Hooks, Complex Rhythms]
[note: emotions: Excitement, Rush, Euphoria]
[note: chords: Gmin, D#maj, Bb, F]
[note: Vocals: Optional. If used, they should be high-energy and distinctive.]
[note: Vocal prompts: energetic and intense, catchy and engaging, fast-paced and rhythmic]
what was changed?
read it
also do I have to include something else, or is just pasting this enough to get what i want
you have to desighn the whole song around what music you want
wdym by that?
wait how would I generate instrumental for this?
oh it gives something closer
here some gens for yuh of the above things i changed its goona be a min to gen lol
lyrics have a major impact on bpms
check it out... brbr
[note: Song details and metadata follow]
[note: tempo: min: 190, max: 240]
[note: bar length: 4/4]
[note: tonic: G minor]
[note: tuning: E A D G B e, 440Hz]
[note: genres: Speedcore, Hardcore, Melodic Hardcore]
[note: styles: Catchy, Complex, Energetic]
[note: moods: Intense, Thrilling, Uplifting]
[note: arrangements: High-Speed, Complex Rhythms, Catchy Melodies]
[note: instrumentation: Heavy Synths, Distorted Bass, Fast Drum Machines, Energetic Leads]
[note: production: High-Quality Recording, Bright and Punchy, Complex Layering]
[note: structure: Intro, Verse, Pre-Chorus, Chorus, Drop, Breakdown, Verse, Chorus, Bridge, Drop, Outro]
[note: dynamics: Intense and Fast-Paced, Catchy Hooks, Complex Rhythms]
[note: emotions: Excitement, Rush, Euphoria]
[note: chords: Gmin, D#maj, Bb, F]
[note: Vocals: Optional. If used, they should be high-energy and distinctive.]
[note: Vocal prompts: energetic and intense, catchy and engaging, fast-paced and rhythmic]
[Intro]
(Sample Speech)
"Fast and furious, the rhythm's calling, let's lose control..."
[Intro]
(Sample Speech)
"Fast and furious, the rhythm's calling, let's lose control..."
[Verse 1]
At 220 BPM, we feel the rush,
The world blurs by in a thrilling crush.
Distorted bass and fast drum beats,
Our hearts sync with these electrifying feats.
[Pre-Chorus]
In this high-speed chase, we find our stride,
With bright and punchy sounds, we ride the tide.
The catchy melodies pull us in,
In this complex rhythm, let the journey begin.
[Chorus]
Fast beats, can't stop, intense and raw,
We're uplifted, breaking every law.
In this moment of euphoria, we reach new heights,
240 BPM, we ignite the night.
[Drop]
(Instrumental Break with heavy synths, energetic leads)
"Feel the power surge, the ultimate release!"
[Breakdown]
(Screamed/Chanted)
"Speed is all we need, let it set you free!"
"In the hardcore beats, we find our creed!"
[Verse 2]
The catchy hooks keep us moving fast,
In this melodic hardcore, we're built to last.
Bright and complex, the layers intertwine,
Each note, each beat, we're feeling divine.
[Pre-Chorus]
The intensity builds, we can't slow down,
In this energetic sound, we wear the crown.
With complex rhythms, we twist and turn,
In the fire of the beat, we crash and burn.
[Chorus]
Fast beats, can't stop, intense and raw,
We're uplifted, breaking every law.
In this moment of euphoria, we reach new heights,
240 BPM, we ignite the night.
[Bridge]
(Softer, more melodic)
In the calm before the storm, we breathe,
The euphoria builds, a spark beneath.
Complex layering, a story unfolds,
In this high-speed dance, we take control.
[Drop]
(Instrumental Explosion)
"Euphoria unleashed, feel the energy!"
"Gmin key, we set ourselves free!"
[Outro]
(Sample Speech/Chant)
"Fast and furious, let the beat take control,
We live for the rush, we burn like a soul."
lol
Welcome to Suno. You do not directly control it, all this stuff is suggestions
and Suno doesn't do exact instructions like this.
Suno is 'vibes', not programming.
Suno thinks this he doesn't want to do any of that
yuh well if its not programed with speedcore genera then its not going to do it
this is what it supports https://discord.com/channels/1069381916492562582/1154790975240155247
Hello everyone, hope you are great! I have a question: I am working on a song, and I want the chorus to repeat exactly the same way the second time I use it... But I am unable to make it sound like I want it to.
Thing is that the second time comes not immediately after, but following some other sections in between.... What can I do to get the result?? Have tried using all the meta-tags but to no avail 😦
Hope you can help me... I am kind of stuck UnU
Thank you in advance
I need help if its possible . Can you add other fonts too like Kannada and Hindi for the video output , and one more thing is text get cropped in portrait vedio so can we get a landscape view video too?
Those are just common genres/styles that users voted to have as discussion channels/threads here on this server; completely separate from what Suno can actually do.
Hello, I am unsure if this question has been answered before, but I'd like to go through the proper steps and know before jumping into anything.
As a paid subscriber to the Suno software, am I able to use my license(commercial rights) with the songs I create and make them for others to use in their own projects to profit with(essentially sublicense)? Or would I just be able to make them for others for personal use.
Thanks for the help and advisement, in advance. 🙂
Are there prompts to help control vocal inflections at certain parts of the song? For example. having them go higher or lower on a certain line, or hitting a falsetto note.
Dude, we just talked abut the tempo thing last night -- I thought you agreed that we aren't able to choose the specific bpm?
i dont give up
You JUST said that it doesn't work, any now you're actively telling people that it DOES work.
lol
in lyrics it dosnt
but instrumental maybe
but your ableto suggest it to the ai with generas
This is legitimately harmful for new users trying to figure things out.
I understand that you're trying to help, but the ways that you're suggesting control over Suno's output is misrepresentative of what's actually possible.
Woodhouse, I miss the days when you trusted me to do shit in here.
Please apply this determination to proving your claims in isolation, instead of persistently repeating things we collaboratively disproved. 🙏
go nuts
this is from yester till to day
all thino i imputed to try diffrent things
That's great -- Is this what's adding the "tempo 220 bpm" thing to your songs?
tbh ... lots of that conversation just reads like chatgpt is purely trying to guess what it thinks would be a good idea based on your input without actually knowing what or how it would work, to satisfy your question instead of actually providing a working solution.
ik i probably come across as extremely dismissive when it comes to metatags, and can some metatags help push suno in a particular direction? yes, sure, sometimes i try and use a couple descriptive ones... but just bcos chatgpt is telling you how to do something just to answer your question doesnt mean that 90% of it is working like some kind of suno black magic. i've not rlly had any luck trying to follow that.
i dont think in its current state that the kinds of suggestions it has are suitable or fully understood by suno and i think if something coincidentally does work in one or two generations, then its going to be just that ... a coincidence from the generation which isnt reliably reproducable (in my experience)
Never rely solely on what an AI tells you.
Also, I'd like to point out that you can ask a custom GPT for the custom instructions that it has been trained on (in this case "SUNO.AI - Helper" ) and they are extremely vague and do not specifically detail any of the extensive metatags used in the above result (it does say "such as [laughter], [whispering], or [rock music]" as well as a list of similar metatags, so anything regarding tempo, key, chords, etc seems to be a total guess based off of its understanding of how metatags (like the examples it has) works, and a pure guess on how Suno might read it, just so it can provide an answer to your question.
too long, didn't read, chatGPT made EDM banger-- i_dunno_man.gif
I'm just seeing all of this from last night, but please reach out to one of the guides or myself in the morning if you'd like a debriefing on all of the misinformation shared last night regarding what you're able to 100% guaranteed "code" or "command" into Suno.
If it works for you and you're happy, disregard, but there seemed to have been a LOT of stuff made up on the spot in attempts to help you last night :/
By chance, was it this issue? https://discord.com/channels/1069381916492562582/1259902708765167709
If you have 1248+ characters input into the lyrics box using v3.5 and hit "generate lyrics" it gives an error -- the current workaround is to use <1248 characters, or make lyrics elsewhere and paste them (https://www.chatgpt.com, https://copilot.microsoft.com/)
To save anyone the trouble, I've tried and am willing to clearly state it does not work
[note generate 5 versions]
Is this intending to command Suno to make 5 different song entries, or is this supposed to separate the 4 minute generation track into separate 48s segments?
We choose to [give Suno a DAW-like interface]. We choose to [give Suno a DAW-like interface] in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills...
#Suno4Prez
A Banger for every Sunovator, and a robot to make garage 🎵 🔊
Hello again guys, sorry for being pushy but, are you able to help me with this? I would be extremely grateful.
Politely following up 6 hours later is the absolutely furthest thing from pushy ❤️ Reading now
Structure tags like [chorus] [verse] etc, can help suggest to Suno that the upcoming section of lyrics is supposed to "change theme", and while giving those suggestions can "nudge" Suno into a direction, they don't act as "codes" or "commands" to explicitly tell it what to do.
When you're extending from a song, Suno can only use the part that it's extending from as "memory", so unless you're extending from "get whole song", it might have to make best guesses as to how the chorus should sound, but if you "get whole song" before extending, it should be able to look further back into the song to see what the previous chorus sounded like.
#📚┃suno-school message Here's a technique I call "Terminator Extensions" that show how extensions can be used to get the exact opposite effect -- essentially cutting the "memory" down to 1 single second so you can stop on a dime and go from opera to heavy metal in 1 second flat 🤣
Obviously you're not trying to make that big of a change, you're trying to add MORE context to the memory, but I'm hoping it might shine light from a different angle on how extensions work to help use them effectively
I tend to ramble in my explanations, so let me know if I can clarify anything for you -- or pop over to the pool party if you want to look at some of my earlier experiments or have any ideas for me to research next 
I wouldn't mind a quick explanation yeah
You are the man bro, thanks for all info!
WIll def try that!
@dire rose yo ☕ ❤️
I hate feeling like a buzzkill, so I want to preface that the absolute last thing I want to do is stop people from experimenting or trying new things.
I've seen a lot of prompt stuff shared that looks like "coding", or specifically telling Suno what to do -- the only problem is that I haven't been able to replicate some of the more specific stuff, and no one has been able to explain it to me.
I might be a little jaded when it comes to the realistic applications of public-facing AI, but I had early access to Bing Chat back when there was only 4 message exchanges before they reset the entire conversation so you didn't overwhelm it -- I've experimented with trying to add contextual & persistent memory to public-facing AI, and while a lot of the experiments didn't pan out, I always learned something from them -- and knowing the limitations at the time didn't mean that the software was any less life-changing, it just meant that I was now better able to focus my efforts.
just piggybacking to say this is defo the way to go!! - almost always use structures like [Chorus] etc. Just imagine youre writing your song for Genius.com as that's how they always do it lol. If you have a repeated part of the song like a chorus but do want different instrumentation, labelling with [Chorus 1] and then [Chorus 2] can help signify they should be slightly different too!
Thanks pal! You're both legends
I was more or less expecting this tbh
For instance, ask chatgpt how many circles are in this picture, and how many are in the rectangle
It won't get it im almost positive
You're able to talk to it like a human, and just because it understands how it should respond doesn't mean that it understands 100% of what you're saying.
Same thing when people give it hyper-specific command prompts.
When I am lazy, I just use this prompt on ChatGPT:
add comments, like [Chorus], [Verse 1]
and then paste the lyrics below
🙈
Sure, it might understand roughly what you're trying to suggest -- like "bpm" clearly indicates that the song should "go fast", and I've seen people getting songs that continue to speed up as they go on, just because they spammed " bpm 200, bpm 200, bpm 200, bpm 200", but it doesn't mean that's the bpm it sped up to.
Hell, that's not a bad idea -- even Suno generated lyrics include basic structure metatags to suggest that the [chorus] should sound more important than [verse], and [bridge] should follow similar "breakdown" kinda rules
The only "downside" comes when people get pissed off because "they wrote [verse 3] clearly, and it sang [verse 2]!" without realizing that Suno is still just a music-robot with ChatGPT hooked up so it knows how to speak.
@unique stone @naive quail, sorry I had to say the B word, I'm just trying to be helpful 😦
i love this prompt even more
translate lyrics to _ somelanguage _ but keep comments like [Chorus], [Verse 1] in English 🙈
Well i know for starters suno modified everything anyways so i was rlly suspicious about this weird copypasting metatag thing in the lyrics
I'm happy w the one I originally made tbh
haah i might have to start doing that to speed through the reams of album-long lyrics i have saved 💡
it usually do god job, but you can always fix errors manually later (or with another prompt(s)
i think theyve fixed it again now but a short while ago, it would only always generate in the order V1, V2, C, V3, Bridge, C for me which would end up messing the whole structure of the generated audio up most of the time 😆 it was always like "really? you want the bridge ... there?!"
I've been made really aware of my own confirmation biases after playing with Suno -- makes me think of the Barnum & Bailey effect, named after the dudes from the circus -- they threw so much at you, that you didn't think about the things that didn't impress you, as long as you came away remembering the things you liked. Don't like the jugglers? There's clowns! Don't like them? What about the lion tamers, or tightrope walkers, or popcorn, or dancing bears, etc.
Similarly, when you throw a whole HEAP of styles and metatags without really understanding how they work by themselves -- how can you be sure which ones had an effect and whichs ones didn't? Is it just a matter that the song sounds good, and you'd be happy with it anyway? That sure as heck was the case for me!
So to stop people from "wasting credits" trying to familiarize themselves with things, I started to make prompt experiment playlists in the pool party #1256301881966198805 -- I've already disproven the bpm thing just last night (with stratuss' help...) so let me know if there's anything else you want me to see ❤️
i just got
[Intro]
[Verse 1]
[Chorus]
[Verse 2]
[Chorus]
[Verse 3]
[Chorus]
this is correct for that song
this is my completely personal viewpoint of how i see it - no one knows how it works for sure - but, bcos that box asks for 'lyrics', i presume its being fed to the AI as "these are the lyrics for the song". so whilst some light and simple prompting can and doeswork, the box specifically isnt dictating the entirety or style of the song... so why would it make sense to write 30 lines of 'detail' in there and expect that to direct the full song
sorry to ask this again, but was there a link on how tyo write timing signatures or beats?
creds renewed, time to dive into this rabbit hole.
That's what I was expecting but I didn't wanna contradict the stratuss guy and the rest and look like an idiot LMAOO
To me it's not that deep I used suno enough to know metatags only take you so far, basic ones can work and that's about it
The rest is so randomized
NONE of us are experts here -- question everything -- especially that weird harber dude.
AI is an unexplored landscape there are no wrong answers
Oh no i'm just saying this based on my own experience dw
Well yes but there are dead ends
Yea that guy knows everything, he used to be blue but dunno what happened to him
there is only a feint direction to help you begin to explore.
Beans?
Bruh
i was doing some experimenting last night on this. the answer is no (not in an accurate and replicable way).
HOWEVER i found that writing 'odd time signature' (or unconventional) as a style seems to sway the AI in a 3/4 or 6/8 timing over 4/4. (i was mostly testing with pop, electronic and ballad styles). ⚠️ this will not work every time and is merely a suggestion based on my own experience which needs more testing.
writing the time signature itself (eg 3/4) is not effective also in my own experience
@stiff sable I think this is the answer to your question, why chatgpt gets better in getting song structure based on lyrics 🙃
identify yourself
ChatGPT
I’m ChatGPT, an AI language model created by OpenAI. I’m here to help with questions, provide information, and assist with various tasks. How can I assist you today?
which one are you 3.5 or 4o mini
ChatGPT
I’m based on the GPT-4 architecture. If you have specific questions or need assistance, feel free to ask!
@charred merlin @stiff sable Wanted to text real quick to let you know your tips made the trick!
is that in brackets? or are there command prompts to write in a specific way?
everything outside of the bracket says what I write.
style prompt for the style of music, not for the lyrics box
Borscht
I didn't take it in a bad way, it's good to question everything that doesn't make sense because that's the only way we learn.
tbh, @untold briar accidentally made me question something that I considered fundamental to my understanding of Suno -- I'm still researching it, but because I knew that we're all experimenting and learning together (instead of him outright telling me that he knows something that I didn't), it didn't make me feel uncomfortable or threatened, it actually excited me to look further into the matter and share my findings one way or the other 💪
yes but in the lyrics box you can direct the song.
[fade in orchestral solo]
[build up synthetic beats]
[building beat and piano tempo]
[beat drop]
[galloping guitar bridge]
[dubstep metal guitar solo]
[beat fade out]
[soft outro]
[fade out]
in the lyrics box to write an instrumental song.
cyber nu post burlesque metal, argent metal, avant garde post country, death piano, arithmetic French folk djent song. Listen and make your own with Suno.
granted I am still experimenting here.
i have a stupid question, but i have to ask it
is there any way to tell Suno what language the lyrics are written in?
if i add lyrics in my native language (Serbian) some things are accented in a totally wrong way, like when foreigners speak my language
I guess I know the answer, but... still had to ask
yes ik we've been talking about that all morning 😁
specifically for what i was saying about trying to get a different time signature using my suggestion, it goes in the style prompt box for it to be somewhat effective
He doesn’t even know how to make the bedsheets, but sure, give him credit for something 😏
i see some people's songs a bunch of dashes and commas and periods is there any literature on that?
like they are writing the beat itself.
this?
@charred merlin Are these rumours true
Exactly, no wrong answers, just experiments that yield results and those that don't 💪
was just about to link that too 🤣
we talked about that yesterday, so it faster to read that
This works great, but the crazy thing is that Suno seems to "understand" when to transition to a chorus even if you don't use those tags -- sometimes an extension can help force it into a chorus or keep it on a verse, but it's wild just how intuitive it is just by itself. v3.5, at least
this is not proven or accurate or reliable or 100% replicable and like everything with suno would be a suggestion to it rather than a concrete way of directing it.
you can try it and see if it works, there's an example on https://www.suno.wiki/faq/metatags/instrumental-tags/ , but this subject originally came up in terms of directing suno on generating instrumentals using non-word text
this message is also very important to consider: #📚┃suno-school message
Also important to remember that a LOT of the community wiki was written based on the understandings of v2.
While a lot of the general methodologies can be applied between models, one of the devs themselves admitted that you need to "relearn" each model, in terms of strengths and weaknesses. Think about how detailed the prompts need to be for Dalle-3 compared to the limitations that earlier models had.
i need to try something that just popped up in my mind
i need to paste some parts of my bash scripts in suno
to see what song i will get 🙈😂
it is more meaningful that random dots and exclamation marks
yess for sure! i dont think ive ever used the wiki myself tbf, i much prefer doing all the experimentilating myself 😉
I know I shouldn't take it so personally, but I genuinely get upset thinking about people who DON'T have the kind of experimental approach that we share -- for instance, if someone tries Suno and doesn't get "the song they heard in their mind's ear", I would hope that they look for the helpers that want to collaborate with them and help them make it -- but I worry that they'll look at super-complicated prompts and feel like the issue wasn't their lack of "play", but simply that they didn't "know the right magic words".
So it's a fine line between trying to let people still be imaginative and feel like they have a degree of control in the magic they create, but also highlighting to others that the degree of control is still within reason
i wish there is a way to get just one song per prompt, so i could try more prompts faster 🙂
It's good to have approachability for all levels of experience and dedication, the people doing it for fun don't have to get too involved in learning how to make something, and the "power users" are here testing the limits (and ultimately, making the experience more enjoyable for them as we "train" the AI.
But yea, there's so much more to Suno.
If I were handed keys to an experimental Ferrari, I wouldn't just look at it n fiddle with it, I wanna see what it can do 😎
what do you mean by "train" 
I agree, but I think I at least understand why they do it that way -- a lot of image-gen stuff will create a "control" image, that way if both images seem somewhat similar, and they both seem somewhat similar to the "expected output", then the AI can be more certain that it's not a weird generation.
In the same sense, it's not like it's being inefficient by making 2 songs, it's like when you order a milkshake at a restaurant and they bring out the fancy glass, but also bring out the steel mixer wih the leftovers -- it costs 10 credits to get the milkshake but instead of just showing you the best result, they're giving you both to work with
That being said, it's a known suggestion (currently closed, so a guide or mod will need to open it for you to upvote it https://discord.com/channels/1069381916492562582/1254418510885687366)
I'm assuming the more in depth users are submitting feedback via bug reports on the site and here, and we're giving thumbs up to what we seem good songs, they gotta be taking that info into account
Question! The stems function would have been terrific, if it was actual stems. Unfortunately it is not a true stem and the music sounds filtered and with artifacts.
Is there a way to fix this? Perhaps run the song in suno again, so it recreates the exact music but without artifacts?
They've confirmed as such that upvoting, downvoting, and reporting bugs are the most effective ways to directly impact future models ❤️
Yup, "vocal remover" is a better name for it, but they've admitted this is only V1 of the feature, which they've managed to squeeze out of v3.5 in late stages.
#🚀┃updates message There's a form they've provided if you want to give detailed feedback ❤️
ohh ok i wasnt sure if you meant the songs themselves are directly training it :D but yes that makese sense
It's really not that complicated nor did I mean to make you infuriated or smth, some people such as me don't have the patience for trying your experimental approach. You shouldn't get mad at that, not everyone is entitled to making progress on learning how the suno AI works. And if they're complaining after? So be it, let them be. If they're that annoyed because they can't make music, they would seek tips at some point and learn, hence why you guys are all here, but doesn't necessarily mean everyone will follow that path.
And on my side, all I did in this message was express my current opinion based on everything I've seen so far. A form of experimental, quite literally, just not the one you want apparently. (Last part unnecessary)
Beyond giving feedback to help them improve, either by reporting poor audio stem separations with ... > report > bugs > ... or using the feedback form in the post I linked to, the best current method for discerning users would be to use a specialized stem splitter, #📚┃suno-school message
so, for now, it's better to use some other software to separate stems...
Check! Thanks 🙂
i dont think it was anything directed purely at you, more so just that some ppl will try and act like theyve cracked the code to suno, using extremely extravagant prompting that is entirely baseless, which isnt then helpful to less experienced people who dont have the time to be testing things like yourself, and makes them suddenly think they might be missing out on a bunch of things by not using those crazy prompts 🙌 that is what is annoying, bcos of some ppl lulling others into thinking they need to be doing so much more to black magic suno into better songs.
ohhh i misunderstood
my apologies @charred merlin
well, assuming someone acts like it is the golden solution, i absolutely agree then
❤️ I wasn't infuriated at all, I apologize if I gave that impression -- I was just explaining why I feel compelled to step in when I see others giving 100% definitive answers on something that we're all still uncertain of
he isnt upset at you, just the idea that some ppl will end up getting tricked into that mindset which isnt actually proven 
personally i just go off the assumption that everything in suno is modified one way or the other
so i try my stuff and if i get cool results then hell yeah
exactly, I was upset for them -- which I realize is my own thing to work on
this is exactly the GOOD kind of experimental approach !
honestly like
not to spit directly at stratuss or whatever, he probably had some good tests with it
but when i saw that gigantic wall of metadata
i was like "there's just no way suno actually listens to all of that"
and well, after implementing it, i got texas cowboy instead of rhythmic speedcore
XD
i did get one interesting generation
but
yeah exactly
i feel like suno just listens to like, 30% of what you'd give him with thiss method
just randomly chosen 30-50% and the rest he just can't process it's too much
giving the most random wacky results but if you're lucky one interesting gen
that's how i felt with this method
and the more data (or therefore 'noise') that you try and give to suno, the more it has to try and pull from and work on which will usually actually just point you further away from what you're trying to do in the first place 😅
@dire rose ❤️ Sorry, I know that I can be standoffish in my approach at times, and while it adding a prefix to the beginning of my rambles doesn't give me blanket-immunity to say whatever I want, I try to start things off by saying that I absolutely don't mean to discourage people from experimenting or trying different things -- if anything, I just want to try to share my experiments so people can try to learn (if they want) instead of throwing their hands up and saying "well, it would cost too much to learn all of the different things I want to try" -- it's a constant battle against confirmation bias, even for myself 💪
yeah
all we have that isn't confirmation bias is the official informations suno gives, when you think about it
technically, most of the rest could or could not work
the "Don't mean to discourage people" harber 😁
for instance, "genre of music" is where you put the genre and it'll try to reproduce that genre
that is something we know for sure and we then try to kind of loophole around everything to help ourselves
honestly I think a dedicated channel in #1252267444773326848 or smth, to explain how the researches are advancing, would be amazing
I'll say, it's also hard to read emotion though text, stuff online get misinterpreted like that all the time.
Pretty sure thats why emojis took off in the first place
loveyouharri
for instance, write smth like :
EXPERIMENT 1 : [Name of the experiment]
- Explain what kind of changes this could induce, or what you're looking to do, are you looking to help your generations be more effective? Are you looking to stop specific bugs? Are you looking to help suno understand better? Are you looking to add something specific and new to the generations? blahblah
And then, detail how the experiment is going, what kind of writings seemed to give what results, perhaps even slightly deeper explanations as to why it'd work?
Or even detail experiments that didn't seem to work on the opposite side, to further enhance the worksheet
and then, everyone can come and give reports, or whatever, and the worksheet itself can grow vastly

something among those lines
it would act as a data saver to everything that has been discussed and seemingly discovered so far with Suno
and enhance the fact that everything is entirely experimental and nothing is a guarantee, to avoid confusion, but I think that sounds pretty sick
well yeah obviously but it is actually not that complicated to read emotion through text, what's complicated is placing everything into context when the texts are too big
that might create a break in the very fine line between emotions and logic, where two thoughts just collide themselves and provoke a little misunderstanding
for instance :
Based even on the amount of action that the #1252257127326548028 channel sees, I'd wager that the people taking time to fill out that feedback form is DRASTICALLY small -- so having well-reasoned insightful feedback from tech-savvy users like yourself is probably an ABSOLUTE GOLDMINE for them.
ohhhh if ppl actually put the time into detailing and working on this, that would be a incredibly fun and interesting experiment
let's suppose im talking with person A :
CASE 1/
me : hey
person A : hey!!
me : how are u?
person A : good! wbu?
me : yeah good too man!
person A : aw i'm glad to hear that!!
CASE 2/
me : hey
person A : hey.
me : how are you?
person A : good.
me : uhh you okay?
person A : ye dw.
CASE 3/
me : hello
person A : hey..
me : wsup g
person A : ehh nothing crazy dw...
me : u ok fam?
person A : yeah.. i'm alr..
so
in this very rudimentary case, but that can actually decline to a ton of online conversations
punctuation, slang, words use, length of answer, ability to communicate, blahblah
everything allows for an educated guess on how the person might be feeling
for instance, it allows us to guess that, in case 1, person A seems happy, in case 2, he seems mad, in case 3, he seems sad, or at least not okay
now, I strongly disagree with all those idiotic "gurus" who proudly proclaim that their incredible methods allow you to read through anyone with only a few tips and tricks (for instance, "Oh if he sends 3 messages only reply with 2! Make him interested!")
sort of what I'm trying to do with the pool party > #1256301881966198805
but, saying that a human is impossible to read is objectively completely false (except if they try to make it that way), humans have emotions and those emotions have logical consequences, decoding it is underlying in basically every single one of us
we can take it too far, we can make mistakes, the other can try to influence us, but emotions are definitely readable through messages
this is also, as you mentioned, much influenced and accentuated by smileys (and also by the topics/tone the conversation takes)
I'll take any opportunity to share @unique stone masterpiece regarding emojis https://suno.com/song/f735599e-cdf9-49dd-aa21-984fe7a9cdbb
dubstep opera, melodic song. Listen and make your own with Suno.
now this isn't perfect at ALL obviously, it's barely easier than trying to decode what Suno will do, but it is only natural to make an educated guess as to how someone is feeling
pfftt
i'll perhaps listen to it later if i can :)
hope that'll work then, try to promote it in that direction
also sorry for the walls of text i just love explaining stuff to people and this community is abnormally nice, so
People at Suno.com,
please implement a routine to batch download all the songs on my profile.
It could be a block of all the songs generated on a specific date or a specific interval of days.
In the 21st century, having to download music one by one is the end of the line.
or if you're him you check the requests sent to download songs in suno
and you copy them in a script to run them following the parameters of each one of your songs (that would probably be spat in a dictionary of tuples or smth)
Never apologize for feeling friendly or being yourself 💙
You're a very welcome addition to the Sunoator Community, and I'm glad you've popped your head into tech-chat 😄
funnily enough i just showed around to, technically, post a piece of work I did that was already existing in chatgpt LMAO
still proud of myself tho
I've heard the idea of batch-downloading suggested before, but the idea of having a filter-specific download request would be VERY helpful -- unfortunately the devs don't often poke their head into this channel for ideas, but if you share the idea in #💡┃suggestions-web or #🚧┃suggestions-layout you'd have a better chance of getting important eyes on your great suggestion 💪
Plus those channels move a little slower, so the idea has more "time in the sunlight"
okay ima dip for now cuz i can't lie
yesterday, with the original thing I created, before all that [note: data] stuff, I generated a song for rhythm games
(instrumental)
Speedcore, Lead Synth, Kick Drum, Vocal chops, Female high vocals, Fastpaced, Chaotic, Intense, Huge drop, FX, Arpeggios
it gave something quite amazing
now there's no vocals so I could remove that but, the result is still rhythmic as hell and damn it's probably the most complex thing I have made for suno
it sounds complex, always changes beat, and still follows a good rhythm, I love it, so I actually wanna put this song in my rhythm games and try to do smth with it
👋
oh wait befor that
like what does this do exactly and do the symbols mean specific things?
so
here's my belief :
there are some differences between instrumental mode, and lyrics mode
Suno seems to understand the both differently
Therefore
suppose we generate 2 songs :
SONG 1/ Instrumental, genre : (insert ur genre)
SONG 2/ Non-instrumental, but there are no lyrics : (insert ur genre)
The two proposed results might be different
because suno understands both as categories that are clearly apart
but, if you try to make a non-instrumental song without lyrics, there is one key difference : you can try to add some metatags to manipulate the instrumental into sounding like what you want
im trying to get at, is there a dictionary that suno uses when you are trying to write the beat itself.
nope
at least none that we know of
suno proposes some random examples, while saying that they "might work"
like do these symbols mean specific things
!
_
.
,
<
?
/
nope
these symbols COULD, if you're lucky, be interpretated as some kind of beat or whatever
that's what suno lets us know
yk it's like
if i write
"p-t k, p-t k, p-t k"
say it out loud
it sounds like ur beat boxing
well
the same way
suno might understand symbols as some kind of beat
how complex could my beat become and I for one am not very musically inclined, I use words more to pull and tug at the concept of the music.
I hella am not musically inclined as well but
i just read these. this isn't the point but i am mildly autistic and i cannot tell the difference between any of these three conversations 😆
o
i'm an asperger
i know what you mean dw
im on the spectrum myself
this is simple social conventions, you can learn them if you can't distinguish them
sorry woodhouse i got excited uhhhhh
lol
i meant hello
thanks for hanging out, hope to see you around more often 😄
social conventions are based on a logical use of emotions, so you can learn them at one point
because they all follow a form of logic
hooray for emotional detachement
😎 i follow that line
but i'm fairly okay in human psychology so
(even if i'm socially a complete idiot LMAO)
so honestly i'm not musically inclined at all but i did manage to get a very interesting beat by just mixing a ton of words, I feel like suno will just generate you something and if you put more words / try some interesting stuff (such as that "Song cut" method that harber gives), you might get complex beats
that rhythm game beat I made, I consider it to be pretty complex for a suno beat
so much variation is going on
take a listen
I am emotionally disconnected. I don't associate that well nor can read emotions, also uh. im honest, like so honest I hurt people's feelings and don't know why.
i get that. tbf i can infer really well irl based on body language and emotion, and ppl cant tell ( i dont think ) when im interacting with them but other than that, i usually just read things with a very neutral approach 🤓
oh boy we're all just living the same life 😅
well I used to be like that
but it's really not that complicated, you learn emotions based on how people act and based on how you feel
people should be happy I say exactly what I am thinking lol
Everything is logical in a way, when it comes to emotions
then, for the honesty part
well i'll take a crude but basic example (very crude tho)
suppose someone's father just passed away
that person should obviously gonna be sad, that is just logical
"i lost my father"
that's not quite my example
where did you last seem him, you'll find him
XD
I know im just recalling a george carlin bit
so, suppose I have to announce to that person
im not empathically-tarded, i just...can't read people.
let's give it 2 ways
way A (the crude, the one you'd use) : your father died (add details you know)
way B (the one most people would use) : hey man, sit down, i have a new to tell you and you're not gonna like it, (insert more blahblah), your father was a really great man, you can talk to me all you want man i'll be here for you, he passed away peacefully in his sleep (or wtv)
I am more in tuned now though, my mom died almost 1 year ago, having had to witness her descent to her death.
oh wow that's harsh
i mean. it is what it is.
my dearest condolences for you still
mom got sick, stopped my life, took care of her for 3 years, then she died.
how old are you?
ohhh
i was mistaking a bit about you then my apologies
i thought you were a person on the younger side that was, as a person with emotional detachment, slowly gonna learn it all
most people online are like that
but you're an older person that has seen it all
i'm guessing ur emotional detachment comes from a lack of care then, after 40 years
well i'll give you a tip, if you can find it, emotions do help in life
even if I agree, simply being yourself is also a simple and quiet way to live
but actually humans are not that complicated, emotions add a layer of complexity and are intrinsically connected to how you'll judge ur life
well that in itself is the start and end of the problem
you're currently half convincing yourself that this is what you're doing
and then, you're half really doing it, because you want it that way
is it as a way of distinguishment, or is it simply how you feel it should be? perhaps a bit of both
the art is a pressure valve for everything that builds up inside.
but, although it works, I won't try to force anything upon you, you're old enough to take ur own decisions
but painting emotions is nothing to be proud of
painting emotions when necessary, that's quite a skill
now gouging on emotions to allow yourself to freely view everything from an omniscient point
that has it's limits
personally i'm at only 18 so
i don't really have the point of view you have in life
and anyways i didn't live your life to begin with
but i found a nice gf and the emotions i've been feeling with her honestly made my life so much better, even if it didn't change that much in theory
now the gf example is irrelevant, but I am fairly sure there is always something new to try
hello al
hi cumulonimbus
lol
for sure my 7 year anniversary is coming up, trying to finish all the commission stuff so I don't have anything to think about art wise./
i'm trying to open this sandwich but i can't
but both of my current clients have me workin on stuff that's labor intensive and time consuming.
seems that I misunderstood what you meant again
a pointilist drawing and a sort of pointilist paintings. both are taking an excess of 100 hours each to complete.
that style looks really interesting
between this and @alpine venture's work, we need an #art channel ❤️
but the stuff i make that is 'me' shit is way more...more
No no no. It is not randomazed. We collectively experiment here, not all ppl of that mine that's all right. When I say "we" I meen the group of funs how loves to experiment and to find out the capabilities of Suno. The basic idea of communication between human and GPT seems to be the same. GPT has extensive knowledge from various areas like basic terms for various instruments, which we did not learn yet, but we can if we are enthusianists for this kind of approach... We are the ppl who are not just experimenting but also learning in paralel. Which means that we are making ourself better by crossing over horizont - if you are not that kind of person that is absolutely fine, but then you will stay in the old way of thinking, like your horizont will be very limited for few hundreds of meters, whilst we are trying to see as far as possible to get the best.
"Me" stuff
i swear suno gave me that as one of my cover arts the other day...
i make most of my cover art in miudjourney for these songs, or mJ + my art stuff
so far most of it is pretty randomized i'd say, it might literally work or not work
i love it tho, its super creative, ive always wanted to pick up painting but i always get sidetracked by something else, maybe some day .... 🙌 i'd be down for a #art channel as a fellow creative, i love seeing what everyone else does
sooo i may look l ike an idiot i gus
this is AI + some of my paintings
Midjourney's older models mix the human touch with the AI pretty well but their current models are kinda terrible at it.
i mean it isn't bad? but it doesn't meet my aesthetic standards
I don't think anyone feels that you're an idiot, we're just really interested in repeatable science-type experiments so we can understand Suno -- it's great to use tools like ChatGPT to get help, but sometimes it disregards stuff that we've already started to understand ❤️
At least I don't think you're an idiot, and I'm happy that you're wanting to help others 💪
if youre talking in terms of your generations and metatagging then, nobody has said that, and thats also not quite the point. we're all experimenting and trying out new things. however, most of our 'metatagging' attempts turn out to be baseless in terms of proof, and cant be replicated for accurate results which means we have to keep experimenting to see what does work!! 😅 let's keep experimenting!
I wonder if there is a "seed locking" mechanism that you can use to really test the tags.
i can only base this off of my experience from midjourney.
k well, play around note: Song Details and Metadata Follow]
[note: Tempo: Min/max 240-999 BPM/BPS]
[note: Time Signature: 10000000/28]
[note: Key: G major]
[note: Tuning: G D G B D G, 440Hz]
[note: Genre: Speedcore,
time sig is intresting
play 4/4 7/8 1/2 2/4 3
but the way you could tell if your tokens had impact was to seedlock your prompt.
21/8
I also noticed the problem with the tempo meta tags. I think it works in more general waz like tempo:allegro.
way to tell is to delete it completly and do 5 gens
this is essentially what we were trying yesterday though, no? i can almost guarantee if i go and try this now, none of these things will be in my output 😅
this is a fairly common suggestion and one i'd love too 🙌 obvs midjourney has it, udio now has it as well so i defo think all of these things combined means its at least on suno's radar. whether we get it is another question! but 🤞 for v4 and beyond.
yes, no problem to add notes.
any ways i spent my credit playig with tempu and time key i could make a play list and share it you can go throu settings a hear no need to watse yours
please send, i'll be interested to take a look when i get off work in 15 mins 😁
But if you have artist and some lovely song, you can ask ChatGPT or LLama to describe the style and techniques in the song or techniques of the author.
the cymbals quality is still low, it is some noise in background...
some prompts dontt work like [admin [load:] does something not sure yet
This is what I really love on Suno. I can do one job in 5pm, and continue next day morning. This is not possible on Udio, because you need to wait 24 hours at least, so when I work in night, I cannot continue in morning.
oh and suno swears alot in some of these i didnt put lyrics
night
What are you expecting it to do?
so im immediately noticing the tempo, time signature and key are not applying
not in a normal way way
some of these kinda slap as songs in general tho lol
i think this is a suggestion from the chatgpt suno bot which we dissected earlier and found that it ... doesnt really actually know anything
re use gens or some thing load othe files
you can also talk to in xml btw
Each generation would be separate and should only take the context of whatever you are extending from into consideration.
just because chatgpt tells you something doesnt mean it actually works
i tested
yes but your playlist you sent the moment ago about tempo/time sig/key was you testing and it didnt work is kind of my point
Why we are not able? I guess Suno determines the tempo based on many factors like arrangment, composition, lyrics, mood and many things. If we'd so great to calculate all these factors, than we would know how to write complete song, lyrics, length of verses, drumming patterns etc... We would have to be experts to find out what is in the "mind" of Suno to get to desired tempo.
and thats good ! you gotta try to see what happens 🙌 keep experimenting. you might come across something awesome some day.
simple answer: because that's just not how it works and its way more complex than that (suno also just doesnt understand music like that bcos its data is essentially just noise.... arranged noise but noise)
admin load doesn't give you any control over Suno. Can you explain where you're basing these ideas on, or is it just what ChatGPT is providing?
well asking what other prompts work for othe general music generating sites ( or ai models ) not just suno.
Because that's simply not the way Suno was trained -- there are different platforms offering training data with bpm and tempo included directly in the training metadata, but I say that it's not currently possible because we've done extensive testing here trying to see whether it's possible or not.
also github , wayback machine to see earlier prompt so on so forth
@sweet wigeon Here's an example of BPM tests that I made specifically trying to achieve "bpm 190", and each attempt failed.
We aren't able to explicitly "code" Suno in the same way that you can't "code" ChatGPT -- it might understand what you're talking about, but it immediately ignores your "code" once it forgets about it
these things are untested and probably won't work with Suno. You need to add metatags to brackets and Suno tends to read/sing some unusual metatags into the song. The AI cannot read this mess.
I don't think you answered my question, or I didn't understand the response.
Can you please share where you're seeing "admin load"?
if its not for suno, its not going to work with suno.
also i'd like to point out the GPT you showed us earlier with a link is trained on documentation which is just this site https://suno-ai.notion.site/Docs-Guides-Resources-38e5ba5856d249a89dcea31655f4fb74 which is based around Chirp v1
sry 184 was the hiest
hey
free version of chatgpt now using 4o mini, and have updated informations, it knows what suno is, but have no idea how to use it, so if you want to use it, you first need to explain "suno rules", with something like this
suno has two prompts, one for lyrics, up to 3000 chars, and the other for genre description up to 120 chars
suno does not know some genres, so you need to mix known genres in that case
Understood! Given the constraints for Suno’s prompts, here’s how you might tailor them to generate a new hypertechno track:
Genre Description Prompt (up to 120 characters):
"High-energy electronic track with fast tempo, driving bass, complex drums, and intense futuristic sounds."
Lyrics Prompt (up to 3000 characters):
"Write lyrics for a high-energy hypertechno track with a dark, immersive vibe. The lyrics should be intense and atmospheric, reflecting themes of adrenaline, futurism, and cosmic exploration. Incorporate elements of rebellion and empowerment. Use vivid imagery and a rhythmic flow that complements fast-paced, pulsating beats. Include a build-up that leads to a powerful, anthemic chorus. Example themes might involve battling through challenges, transcending limitations, or exploring unknown realms."
ok there we go, that's cool, so it works ..... 1 out of 500 times or so on average
lol
one song is c....
other sounds great, but have nothing to do with what i wanted 🙂
i would absolutely be down for tempo selection, i dont want to sound like im dismissing that, but i just dont think currently the model is 1) trained in a way that supports or understands it and 2) is competent enough at understanding metatagging in a reliable and consistent way
this gave me a nice intro in both versions
[Intro]
(Instrumental – Atmospheric synths and futuristic sounds build up)
Yes, but here I am not sure, because I have tried both Suno and Udio, last time Udio. I have created English song, and inserted Czech lyrics later. But with Suno I have experimented too. So I have made song in Czech, Hebrew and Arabic. But the problem there is that it does not rhyme. Which is very big problem if you want to create song.
I've also found accompanying () bracket prompts next to the [] can add to it. Sometimes () get spoken out loud, but If you have it consistently set up, it's worked well for me
But, I'll usually do multiple () for each separate prompt
So like [Guitar Solo] (Wah pedal) (reverb)
suno has no problem in understanding other languages (including nonenglish sounds and letters), only issue is accent
and only in some words
it is usualy great
It was not really metadata, he did not inserted them in brackets, only few. But Suno cannot understand such a mass. That's too much.
but some sound just wrong
no. GPT, Sunno including, interprets tokens in context.
this song is an absolute banger, made in collaboration of ChatGPT 4o and Suno, I was just a middleman, so I think it is appropriate to post a link here (if you disagree, i will remove it from here and post in showcase)
High-energy electronic track with fast tempo, driving bass, complex drums, and intense futuristic sounds song. Listen and make your own with Suno.
prompt for chatgpt was
give me a prompt for suno to generate new hypertechno banger
suno has no idea what new hypertechno is, you need to explain
suno has two prompts one for lyrics, up to 3000 chars, and other for genre description up to 120 chars
those 3
oh, sorry and
create lyrics for me, add comments like [Verse 1]
SUNO.AI - Helper this is the the helper i use its written for suno
just look for it in chatgpt
i am using free version
you can access custom GPTs up in the top left using the 'Explore GPTs' button
it's not 184 bpm, this is "slow" music, is it like 105 bpm?
altho it doesnt rlly understand much more than normal chatgpt considering all that's really different is this
moise says it 184?
it is 184, i counted and put it in a daw
I tried this: [drums tempo: metronome 180 bpm]
[drums pattern: BSTT + fills and licks]
[galloping bass guitar bridge]
[dubstep bass guitar solo]
The build up is pretty fast. https://suno.com/song/2b39007d-2e19-4a0d-b1ff-b353adfac6ed
metronome, drums, experimental song. Listen and make your own with Suno.
We know the tool you're suggesting, we're trying to let you know that it doesn't work the way you seem to think it does.
@tight spade Show this picture to ChatGPT and ask it;
1- how many circles are in the picture
2- how many red circles there are
3 - how many circles there are inside the black rectangle
Hopefully this helps illustrate that just because ChatGPT gives you an answer doesn't mean that it's correct.
(inb4 "chatgpt 4o can do this, so it can obviously do anything)
LLama: "Dubstep is a genre of electronic dance music that originated in South London in the late 1990s. It's characterized by its heavy, bass-driven sound, complex rhythms, and often features elements of UK garage, drum and bass, and hip hop.
Dubstep typically features a slow tempo, often between 100-140 BPM, and emphasizes heavy, sub-bass frequencies. The genre is known for its dark, atmospheric soundscapes and often features complex, syncopated rhythms.
The galloping guitar bridge and dubstep guitar solo you mentioned are not typical characteristics of dubstep, as the genre is primarily electronic and doesn't often feature live instruments like guitars. However, it's possible that you're thinking of a specific song or artist that incorporates guitar elements into their sound."
🧐 So when I used dubstep, Suno could not use the guitar solo and it had to slow down... https://suno.com/song/2b39007d-2e19-4a0d-b1ff-b353adfac6ed
metronome, drums, experimental song. Listen and make your own with Suno.
i guess the answer is similar to "how many letters R are in word strawberry"
The word "strawberry" has 2 letters "R."
i love this one. its so simple and so funny
As in "commanding/coding" Suno to do things, like explicitly telling it what BPM, key, time signature, etc, to use?
Yes
I wrote about it I'm general
letters, not sounds
Got it! In the word "strawberry," there are 2 letters "R."
🙈
"You don't code Suno, you talk to it."
I used an audio file for an experiment.
Well it might be possible to code suno soon
It's called glitches and bugs
Audio Upload and Extend are AMAZING tools, and audio upload is the only 100% guaranteed method to have granular control over Suno's output, because that's the only way you can have granular control over the input
We're a friendly group of nerds, lots of experiments and passionate discussions -- we might look like we bite, but we're really just interested in seeing people "show their work" instead of somene just showing a random cool song they made and saying "I made this, so I know everything I'm talking about."
This is the channel where the "experts" will passionately fight you to explain that they AREN'T experts, and would rather their observations and findings be discussed than themselves ❤️
(or at least that's what I see it as
)
Oh ok
I made an audio file from an old generation
Of eleven labs sound effect
Just a random sound
It sounds like a turkey animal
Oh cool! I tried uploading 60s of me reading the other day, and it quickly dropped my accent and turned it into something else -- I suppose it's easier for Suno to recognize patterns and try to replicate them, while Eleven Labs trains on the specific voices themselves to extrapolate on it, rather than just repeating patterns like a parrot
Warning! Is very repetitive.
Here is my original sound file I used for the experiment
But ElevenLabs is also handling a way different use case
I made it so the weird sound stops after a certain amount of time for the upload sample
Maybe at 35 secounds
Yes suno ai likes to duplicate and clone.
I believe stale reference manipulation can be used because that
The more repetitive the sound and many sounds
The better the output is
On extend
It's almost as if it's like a training board.
Or pattern key hole.
Idk how to explain it
True - getting a rotating evolving song requires a lot of encouragment
I did not remove the link. Wait, I will copy it again and post it or I will create playlist to experiments.
I did make a theory on this ages ago for something different but similar
How ai could use a consistent pattern and changes in between sections
I forgot most of it now
But I based it off of the subconscious how everything we see should be upside down
Maybe ai is the same.
If you move it backwards
It will move forwards
Add a opposite/counter to a sound
The opposites would then create a desire output
That's my theory
I believe it might work
I could make a fully random sound audio file with only a few parts that make sense
The ai could use that to distinguish between the noise and what is not nonsense
And create the output based on my question
Like a calculator
If I ask it something it know then it will tell me
If I give the audio that has the answer in the noise ans ask it for that answer.
It could possibly give me the answer back
That is my new theory
Drums pattern:BSTT means 1st Beat - Bass 2nd beat Snare, 3rd and 4th beat tomtoms - But suno didn't play it because of dubstep is electronic music.
It's like coding except I'm teaching the ai only a snippet of code which is only useful for 1 use
There is 2 possible ways it could be done
1: create a single noise audio file with answer been said once aka the sound you want.
2: repeat the answer a few times inbetween the noise and extend from the answer
With request ofcourse
These may or may not work
I can prompt but I need to think style because dubstem, EDM, house, hip-hop it is not the kind of music where drums and guitar could be used, but I would like to skip synths.
oooh good toys. How much do they cost?
https://suno.com/song/fb276980-0b10-4093-a551-b0ddbccac5f7
Another experiment with extend and prompts, didn't really work but i tried.
[Anime people arguing] [Silence] [quiet] [no music] song. Listen and make your own with Suno.
Were you just trying to get spoken words with this one?
Hey. Can someone explain where to put meta tags. I'm trying to have a song with female chorus and male rapping. I've tried [Rapped Verse], [Verse Rapped], [Male Rapping] and every other possible meta tag before every verse. But if i add a meta tag for female vocal in the corus the whole song has female vocals.
Metatags go in the lyrics section, think of them as "targeted style prompts" for specific sections of the song
Trying to get specifically-gendered vocals is an ongoing struggle, as it's largely deterministic based on the lyrics / metatags / styles
you need to think about them as "suggestions" instead of "instructions"
that makes sense. I just managed to get 1 of the verses male and the rest female even though all verses has male as meta tag.
This is all very helpful, and while being educated on musical terminology and music theory is key continued education in the field, you do recognize that we're still not able to apply this to Suno, yes?
check this
"Metatags within the lyrics might assign multiple voices, but the gender and consistency are not reliable"
[Mary]
This is my line…
[John]
But now I am singing…
[Mary and John]
There is a chance both of us are singing together…
Here we go, another experiment with drums... I call it "progressive rock, metronome db-90, fills and licks with a muffled snare tap"
[timesignature: 8/8]
[drums tempo: metronome 180 bpm]
[drums pattern: BSTTCTT + fills and licks with tap on snare]
[galloping bass guitar bridge]
[bass guitar solo]The guitar plays like 180 bpm but the drums seems slow... And they play cymbals more than expected.
https://suno.com/song/381c04b2-8f0e-402f-98f2-72f8b5d1dd6a
But now we are talking! This looks like 180 bpm for both drums and guitar! Nice!
https://suno.com/song/d0d10f91-3499-47a8-9f68-7dfa091de136
progressive rock, metronome, db-90, experimental, muffled snare tap song. Listen and make your own with Suno.
progressive rock, metronome, db-90, experimental, muffled snare tap song. Listen and make your own with Suno.
normal rock is not that fast. 195 is what? metal?
Suno may interpret db-90 as tempo... may or may not. IDK
You need to think like: Suno will not create music what you want, if the drummer is unable to play that instruction so fast...
You don't see the problem with getting 5 credits back, and hoping that the system makes another mistake so you can actually use those credits?
@bleak light
It costs 10 credits to "generate", the same as it costs 10 dollars to buy a 2-pack of fancy cupcakes
If you buy them and one cupcake is smushed when you buy it, the cashier will give you $5 back
Is the problem that they haven't given you enough money to buy 2 more cupcakes?
Yes, because you can't buy one cupcake
Again I ask, you don't see the problem with getting 5 credits back, and hoping that the system makes another mistake so you can actually use those credits?
It sounds like the problem is that you don't WANT to buy 2 cupcakes, but that's a separate problem.
You will never be able to use the 5 credits you get from a refunded generation unless suno makes another mistake with generation.
#📚┃suno-school message I just decribed the concept of the 2 songs per generation here using a milkshake as an example
You use the 5 credits WITH THE OTHER CREDITS.
It's not wasted, because it goes back into your "credit wallet"
The only way it's "wasted" is if you're left with 5 credits and refuse to buy more
yes?
Except you don't. I'll lay it out for you
You get 50 credits, and go make some songs. One of them is too short, and it refunds you 5 credits. You now have 45 credits. You make 4 more songs, and now you have 5 credits that you can do nothing with.
Oh just buy more credits, sure, I'll subscribe and get the 2500 credits. Now I have 2505 credits and the same problem
Are you getting it yet?
You have to depend on suno to make another mistake to actually use the refunded credit
Ideally, you'd prefer that Suno NEVER makes mistakes, and only ever generates a single song -- without questioning WHY two songs are made.
Right, which is why depending on suno to make another mistake so that your refunded credit isn't useless is a bad idea
I understand what you're saying, but I feel like you're actively trying not to understand anyone else's opinion, because thinking the credits are "wasted" is very shortsighted.
I understand what you are saying, but it feels like you are ignoring what I am saying
Yes you get the 5 dollars back
Yes, you're saying that if you subscribe and get 2500, you now have 2505, and after making 250 generations you'll still be stuck with 5 credits, which isn't enough to buy a 2-pack of cupcakes
Right, and you can never use those 5 credits unless you get 5 more from another mistake
But it shouldn't make mistakes
So relying on something that shouldn't happen is not a good idea
In a perfect world, there will never be smushed cupcakes, and everything would be perfect.
Except we're not in a "perfect" world -- it was JUST invented less than a year ago
Right, but thats irrelevant, the issue isn't that it makes mistakes
The issue is that if you get 5 back from the failed generation, you can't actually regenerate the song with those 5 credits
Even in a store, if you see a 2 pack of cupcakes for 10 dollars
You cannot open the pack and buy one
I understand your idea, I just don't think it's informed on why things are the way they are, and I'm not sure we're able to find a common ground at the moment.
Perhaps you could share your ideas with #💡┃suggestions-web to tell them to re-think the credits idea
To be clear I don't have a problem with them giving 5 credits back for a failed generation
https://discord.com/channels/1069381916492562582/1254418510885687366 or you can upvote this idea
The problem is you cannot use those 5 credits at all
Do you get what I'm saying? There is a "generation failed, credits refunded" but there is no "retry generation" button
If the generation fails, you don't even need to give credits back, just display a regenerate button on the song that failed and try again
It's really not that hard
These are great ideas for #💡┃suggestions-web, but it's not something we can help you with here in the tech-chat ❤️
But you understand what I am saying now right?
I understood what you were saying from the beginning, and again, it seems to stem more from misunderstanding of why things are the way they are.
But yes, I understand. It's something you should share in #💡┃suggestions-web so the devs can see it and consider it ❤️
see that I can get behind, let's say it either fails or is too short, rather than just auto refund, keep the generation there, but an additional button to retry at no cost, if it does fail again, then we can just give up and accept the ai tried to redeem itself
Ok so explain to me why things are set up so that you have to wait until suno has another failed generation to get enough credits to regenerate something
Dude, I don't believe this conversation is in good faith anymore, so I'm going to go do something productive like snuggle my cats, but I shared a link to my milkshake analogy earlier that explain it if you're genuinely interested.
You explained how it works but you haven't explained why it's like that, despite saying that my opinion is formed due to not understanding how it works
I understand the desire for things to be a certain way, but the only magical wishful thinking that matters is what happens inside Suno's brain
Pretty much every public-facing generative music tool I've played with, and a lot of early-access LLMs give you comparative tests to ask which one you prefer.
It's not a wild concept -- I understand it would be great if Suno just KNEW how to make perfect songs every time, hell, if you know how to improve the methods they're using, forget #💡┃suggestions-web and apply directly at https://suno.com/careers
Just in case Vanta doesn't post it in #💡┃suggestions-web I hope someone does, because that's genuinely an idea worth sharing, once we get past the complaints about why things are the way they are, or wanting to generate more/less than 2 songs
@wicked mortar I feel like we've never really chatted, but I always see your name reacting to things ❤️ I hope you're having a good day 🙂
Even if it requires a control generation, it shouldn't be displayed, and I will argue that we should not be paying for said control image.
For people that want one song generation, just turn it off.
If you train an AI to be dependent on a control generation, it will always be dependent on that control generation. Stop buying the alcoholic whiskey and just implement a generation feedback system instead with occasional double generations and asking which one you prefer to improve the generation model
Same to you 😄
Again, I understand why you feel this way, but these really aren't things that we're able to assist with here in tech-chat -- and the devs don't even pop their head in for tips, they go to suggestions/bugs
I'm not sure where you got the idea that I am asking you to implement these things. I am discussing my opinion
I mean, you're replying to me while describing it, and I'm telling you to describe it elsewhere.
For example, let's say that once every 50 songs you get a double generation and it asks you which one you prefer, you pick one, and it refunds you for the one you did not pick (if you have to pay for the 2nd one at all)
But if I wanted to suggest it I would be suggesting it. I'm discussing it
That's what the chat is about is it not?
No, this is a chat about your suggestion, which seems more suited for #💡┃suggestions-web -- this is technical-chat, where we discuss the technical aspects of Suno -- not what they COULD be, but what they ARE
I'm not the boss in any way, so feel free to talk about whatever you want in whichever channel feels appropriate though ❤️
Well then the channel should be renamed to "how-it-works"
Sounds like a great post for#1252257127326548028 🙏
The mods set that forum up so you can request all of the specific changes you'd like to see in the discord -- like what channels you want, and what you'd prefer that they be named 💪
Suggestions browser is designated for feedback, people should not be having conversations in there, if you want to be technical about this channel then be technical about the others
I'd love to have a discussion about what the different channels' purposes are, maybe instead of a channel dedicated to technical questions we could make a post in #1252257127326548028 and discuss the concept further in the comment section of that post?
That way we're at least trying to organize our conversations ❤️
A modern and innovative logo for Wafra Smart Solutions featuring a unique main icon, consisting of three vertical columns of varying heights and a curved element on the left. The curved element starts from the bottom left, seamlessly connecting to the first column. The right column is the longest and features a small arrow pointing upwards at the top of the column.
Add the Start Engine icon after the last column to create a professional and distinctive look that signifies both a start and an ascent.
The main icon uses a gradient from dark purple to pink, creating a stunning visual effect. The word “Wafra” is placed directly below the icon, using a modern and bold sans serif font. A gradient from dark blue to light blue has been applied to enhance readability against the white background. The phrase “Smart Solutions” is placed at the bottom, using a thinner and more elegant sans serif font in purple. This design ensures a clean and professional look while maintaining simplicity and a distinctive style. The logo is designed with multi-colored lines.
I think you might have intended to post this in a MidJourney server?
@charred merlin I was looking for a way to get less robotic voice in musics ? I saw your conversation and was curious to know if you discovered anything new ?
how is it?
Probably not the expected results... https://suno.com/song/19b330db-9cd8-456e-b41c-c3f32baf25b4 
Dark Horror Poetry, Spoken Word, Glitch Audiobook song. Listen and make your own with Suno.
Admittedly I haven't experienced too many voices that I would describe as "robotic", could you share an example of a song you've made that has a particular robotic vocal tone so we could look into it and try to see what's going wrong?
https://suno.com/song/114df348-87b3-416d-b51b-e5465229d345 this one from times to times is quite robotic
I have some others but most of them tend to have that robotic / autotuned tones that's not pleasing to hear
I'm trying to generate lyrics, and it's refusing. I used to think this would happen because of naughty words or being over 1000 words, but I can't figure out what could possibly be wrong with this paragraph in 'generate lyrics'. Please Help.
"Bloomburrow
is the first set in the Dragonstorm Arc. A ragtag team of anthropomorphic adventurers must band together to go on an important quest to save the realm from elemental calamity beasts and the Great-Night Owl. Glarb, the King of Frogfolk and a powerful wizard, is said to have insights about the calamity beasts. It is now up to Mabel and her colorful companions Gev, Hugs, Finneas, and Zoraline to redeem their world.
The animals have an advanced culture and are the main characters of the set. They wear clothing, talk, and wield magic. Unlike humanoid "animal people" on other planes, the animalfolk of Bloomburrow are the size of regular animals of their species. Much like Segovia, foreign Planeswalkers are changed when 'walking to here; in Bloomburrow's case, human Planeswalkers are transformed into anthropomorphic animals like the local races while here. It's unknown if traveling here by Omenpath will transform oneself, or if only humans are affected."
https://discord.com/channels/1069381916492562582/1259902708765167709 is it this error, or what does the error message say?
this one https://suno.com/song/0f108b35-209a-4941-80a9-071e83d62001 is tough asw
J-Rock, Metal, Hard Rock, Cute, Upbeat, Energetic, Female Vocals, Japanese song. Listen and make your own with Suno.
Is there a particular section of the song where you think it sounds particularly "bad"?
Not quite sure particularly what would be bad about it, but you could ask ChatGPT what would be considered inappropriate since I think it uses ChatGPT as a backend for the LLM stuff -- or you could generate the lyrics in chatgpt and paste them into suno?
for the last one I shared, from 10 to 15 let's say and 1m16 to 1m24
Thank you, looks like I can generate lyrics in Chat GPT. I've never used it before for anything. Wish I didn't have to leave the Suno ecosystem.
I know sometimes Suno can hit specific notes "awkwardly", but I'm not quite sure I'm hearing those particular differences.
That being said, adding a metatag before the lyrics can help suggest specific styles -- not an exhaustive list, but some ideas to try https://www.suno.wiki/faq/metatags/voice-tags/
oh ty for the wiki
@@charred merlin[Pre-Chorus] Chat gpt used this tag, will Suno recognise it?
I was looking for that
#📚┃suno-school message Yeah, there's lots of different structure metatags you can use, but the important thing to always keep in mind is that NOTHING is a command/code to Suno at the moment, it's all "suggestions"
seriously woodhouse, I'm trying to help people, you're usually fun company, but it's getting frustrating.
So saying [pre-chorus] might act as a suggestion to suno that the block of lyrics right after it should be delivered as a prechorus, but that's assuming it's not overpowered by the other metatags, style prompts, etc
Not a problem! #1252267444773326848 has a bunch of user-submitted guides and tips, including the pool party where I post "prompt experiment playlists" when I remember to make them > #1256301881966198805
yeah I kind of read it earlier on
@charred merlinI'm wanting to get chipmunk voices, this style tag "Chipmunk vocals" doesn't seem to work, what should I put?
hrm, so let's try to imagine that Suno is someone from a different culture -- we understand Chipmunks to be the high pitched silly recordings, but if they never had that series in Suno's culture, they would have no idea what we're talking about
🐿️ ❓ 
So let's try to think of different ways to describe the same kind of vocal effect -- high-pitched, squeaky, sped-up
I'll try high-pitch vocals
Maybe we could try "falsetto"
thanks never heard that word before, learning a lot of new music culture words
falsetto is like when you sing super high, higher than you normlly would be able to properly "project" your voice
😄 It's really exciting how many new things Suno introduces people to -- I was a live musician for a long time, and I never personally got the chance to "mix and master" the songs, I just sang and played guitar -- as much as I felt connected to the music because I wrote it, after playing my part I had to "hand it over" to an audio engineer to fiddle around with until they gave us the finished record -- and that was that.
Now I'm being introduced to SO many tools that are free on the internet (probably even better than what the old record studios had and charged butt-loads for) #📚┃suno-school message , and I'm able to take part in every step of the music industry -- music creation, advanced audio editing, mastering, mixing, and even releasing into stores -- heck, my friend @mystic flume even got her music put on CD!
Welp, went with a deeper then ever so far male voice 😦
Oh man, Suno is fun😛 chances are "falsetto" is being overpowered by something else, but if you want I can take a couple minutes to see if I can figure something out? You can share the song here or DM it to me ❤️
shit's broked, guys...
https://suno.com/song/78814a81-6d9c-4edf-ada0-6f8c530a7da8
To be honest, the output is the kind of thing I'd expect from that kind of style prompt 😄
Tums in aisle 7 got me pretty good
This... this is something else... this isn't based on an audio upload at all?
It's more or less exactly what I hear in my head when I walk into a room and instantly forget why I left my office in the first place.
No its public, its all just from that prompt
That is totally mad. It sounds like several pieces of music are playing at the same time
I thought I'd forgotten to close the tab with the other song in it
lol at 01:05 it suddenly restarts
It sounds like a radio phasing in and out of tons of channels at once, I heard Chipmunks, Punk, Pop, Dance....
There's a beat in there somewhere...
I tried a couple v3 and v3.5 generations with the same prompt -- wasn't able to get anything quite like it. Very peculiar for sure -- I'd report those as ...>report>bug>poor audio because something weird certainly is going on 
audio uploads can be made public, and even clips with discernable vocals are able to be made public if the detection system doesn't catch them for one reason or another.
Either way, that still is wild.
yeah the chipmunk vocals were interesting
im so confused. joined this server to try and use the bot to produce a clip of a song i wrote so that i can base my own version off of it, but i dont see the "chirp" bot that the youtube video showed. any help as to how i actually access the AI?
Hi there!
If you’re ready to dive into creating music, head over to Suno.com.
For more detailed information, check out docs-guides-resources.
If you have any more questions, just ask in #general-suno where you can always find our community, moderators and guides to help you with most questions. 😊
Chirp bot has been retired -- video is outdated.
Hi Harber ❤️
Hey Walnewt 
Thank you!!!! Sorry for asking as I am sure you get many people asking the exact same question
It's all good 🙂 We usually have people directly trying to use /Chirp without any typical followups (by the person invoking, Woodhouse set up to respond about heading to suno.com)
Not a problem in the SLIGHTEST -- for the longest that, that was the only way to make music -- there was Chirp for music and Bark for text-2-speech.
After Chirp really took off they spun it off into Suno and open-sourced Bark so anyone can run it on their computer or on the web 👀 https://github.com/suno-ai/bark
Haha I gotcha. Yeah I am not looking to really use whatever AI makes as a "draft" so to speak so that I can write and record my songs, so when I saw the video I was very interested
A lot of people might have older documentation (like old youtube videos, or stuff translated for their languages that arent updated as often), so it's completely fair that THIS IS the place where you used to use the chirp command.
Now that you're here though, the fun really starts 😄
So to clarify you're wanting to use it more seriously, or just to play around?
@teal raft check this out... interesting results from the given prompt
Just play around. I have written and performed many songs, but I am actually working on a new album that is a completely different type of music and have no idea as to the tempo/vibes of the songs i have written, so i figured some type of AI could point me in the right direction after hearing it
I recommend checking out the wiki (linked at #1151542418991239168 with other useful guides) first then try out the Custom Mode feature with the topics you learned about.
It will make your experience much better than what may be generated in simple mode (which has nothing wrong but custom mode is a lot more flexible in trying to get something you had envisioned)
Thank you!
this is how i love to use suno too (most of the time)! super fun to hear your song concepts translated into instant ideas! hope you end up having lots of fun with it 😁
That being said, do not be afraid to try out simple mode too, especially with the adlib-like templates that very recently came out. The simple nature of simple mode can produce some unexpected gems.
A few beginner notes:
- As a free plan member, you can submit up to 5x a day* as each generations costs 10 credits (generates 2 songs per submission) and you get 50 daily.
- Credits renews, it does not get carried over (excluding top-up and gift credits).
*The real limit is the amount of credits you have for generation. You get 50 credits per day which resets at 0300 UTC+0.
Refreshes at 0300 UTC+0 (<t:1713236400:t> in your local time)
^ looove using simple mode sometimes when i randomly feel like it, often it creates some super great results i wouldnt have imagined with how to prompt in custom. no one knows the AI better than the AI itself ;)
though some ppl will try to convince you otherwise 🤭
was trying to get some cool dubstep music with suno, but it always goes into some kind of pop music. what could be a style prompt for such a music style? https://www.youtube.com/watch?v=p2hPUbnHc94&ab_channel=DubstepuNk
Crankdat - Pepper Spray Step
→ Available now! Listen everywhere!
🌎 https://laylo.com/crankdat/REMIXMACHINE
Subscribe & Turn on notifications 🛎️
http://bit.ly/DubstepuNkSubscribe?sub_confirmation=1
🟢 Spotify: http://DubstepuNk.uk/Spotify
Follow CrankDat
Instagram: https://instagram.com/crankdat
Twitter: https://twitter.com/crankdat
SoundClou...
Usually if I have trouble invoking a certain genre in 3.5, I make the initial track in 3.0, which seemingly has a better understanding of various genres than 3.5, then extend into 3.5 for the song structure.
Forgot to mention, one version had "You've Got Crabs" in the style prompts, the singers would sing that line a few times in the intro
i would recommend v3 over v3.5 for that kind of thing, however i've found for that kind of thing, prompts like:
wobbles, heavy, gritty, wobble house, dark, big room, phonk, electro house, intense energetic, dubstep, edm dubstep, riddim work quite well (some of these overpower others so asking for 'wobble house' or 'electro house' gives you the vibes you want whilst keeping it in a dubstep tempo and style (im not saying use all of these at once for a single generation))
some of my experiments you could maybe see for a starting point: https://suno.com/song/9d390d83-d797-45be-92b6-93f76e332b5a v3
https://suno.com/song/082053c4-d6e6-445e-8c4a-485328708699 v3.5
(i didn't even outright prompt for dubstep in these but its close)
is there any way to export lyrics as srt or something to be used in other software?
Unfortunately not yet
https://suno.com/song/c9cc7829-5e3b-4c0f-b73b-c1e1ca7c4311 Im sorry if this is not the correct chat to share, but I am in awe. This is almost exactly what I envisioned for this song
Sad country Style, Acoustic Guitar and Piano only song. Listen and make your own with Suno.
There are some services that can estimate the timing of the lyrics when you use them without timings, I know YouTube can do it since I used to do it manually
hehe i recognise that voice from trying to experiment with country styles the other night
im so glad that youre happy with your result, i love that feeling 😁
i really love those lyrics btw, very heartfelt and the way they tell the story of the song is really nice ✨
Is there a way I can get that exact instrumental?
Found it
https://suno.com/song/c9e69828-d58f-4526-aecd-3b562cc39faf
And this one, they waved in and out of saying the first line and the last style prompt. It's so weird
https://suno.com/song/b222424e-4f92-4518-8833-f024c2984a33
Thank you!!!! I have always been very lyrical, unfortunately was not born with the singing talent to go along with it
im not sure if the free plan has it yet, but you should be able to click on the three dots next to the song and do 'Get Stems' to seperate vox and inst (but its very early days and you might be better off downloading the song as a wav and uploading it to a service like Bandlab Splitter https://www.bandlab.com/splitter to get the seperate parts
If you mean by generating pure instrumental tracks, there is an instrumental switch option, though I personally do not like using it and opt to use instrumental tags instead (more control and less hallucination)
Shiroken had noted (https://discord.com/channels/1069381916492562582/1261146873331318835) that three metatags can be used consecutively without a break between them
Example:
This one just... Ignored the lyrics
https://suno.com/song/43d2a4d5-7d0d-42c1-907d-7f9321c88ad6
it is a cool instrumental tho 
^ I hate and love it when Suno does that. At least it is an uncommon or rare occurrence, though it depends on the genre too I find
(Sorry for the ping, forgot to turn it off in the reply)
np 👌 sorry I keep popping in with test stuff
So long as it is contributes to the topics on hand rather than being used as a showcase alternative, it is fine
Guys! Cannot find any way to do it, maybe you have a solution?
I want use simple mode and generate 20 songs with 10 random lyrics but in EXACT style I define, for example “jazz” (or any specific multi-word style that I prepared and tested before), but with no any additional/removed style tags. Every time Suno adds some tags or removes it from style description.
Did you find any prompt to do it, to prevent from changing defined style? Testing everything.
Only way as for now is to open 10 tabs in browser, copy-paste style, use custom mode and generate 10 texts, then press generate tab by tab. But it takes soooo much more time, better to use it on listening and finding best ones.
Anyone tested such prompt in simple mode and succeeded?
@naive quail has gotten style bleed through a few times, and it’s always hilarious — “Enraged Goats on Welfare!” being screamed at the top of someone’s lungs is QUITE a way to start a metal song off 😅
This is the issue with simple mode — you’re looking for the type of control that “custom mode” gives you ❤️
“It’s a feature, not a bug”
I know, but no random lyrics 😂 need to put new lyrics all the time.
In custom mode, you can type whatever you want and press “generate lyrics” to create the same kind of lyrics, except you’ll have EVEN MORE control over what you ask it to generate 👀
This way you can get the exact same lyrics for 20 songs, or choose different lyrics for each pair 👀 💪
I like using rateyourmusic.com to look up genres or songs (the latter to see what styles they used).
If you scroll to the bottom of the genre page (hierarchy section), you can see all the subgenres related to said style, which is very useful for Suno generations for the style you want.
E.g. https://rateyourmusic.com/genre/dubstep/
I tried testing some dubstep entries which you can see in the new part of my profile.
You know... I just remembered we had this discussion recently regarding multiple free accounts, after seeing the name of the Suno account you were posting your crazy prompt experiments from... I count 3 accounts you've been sharing songs from recently.
Thanks @charred merlin, got no any solution. Hoped it might be in future bug nor feature. 🙂 I’ll ask guys later, maybe anyone found any prompt solution.
has anyone found a way to get suno to make instrumentals that loop somewhat seamlessly?
like if on loop the beginning would run into the end without too much of a jolt
If I understand, you want to use Simple Mode, but you don’t want it to add extra style prompts, correct?
If so, it’s not something that can be prompted differently — Simple Mode is designed this way so people can get different results without needing to know different musical terms
So every song doesn’t sound “so similar”
Did you test my old Suno Composer? Very helpful, included rewriting existing songs (I mean syllable count and topic, not words).
https://chatgpt.com/g/g-ak1zo6jRu-suno-song-composer-by-michalai
To be honest, I’ve only seen ChatGPT cause more confusion than good — specifically relating to Suno.
Unless it’s specifically used to write lyrics, it always leads to more misunderstandings — I see this GPT is geared toward syllable count and flow, which could be helpful, though! 💪
I’d say the easiest way to “guarantee” it would be to upload the looping audio and extend from that — 60s if you could that way you’d have a solid musical context to “hook onto”
Here’s an AMAZING suite of free web-based audio tools if you need to separate stems, edit, loop, ANYTHING 😄
thank you, i'll look into that, it isnt overly important only I'm making instrumentals for dnd games and it can be distracting if the dramatic fight music suddenly restarts
If you’re anything like me, it’s not important until you get 80% of the way there, and you get determined enough / attached to a cool part in a song, and you might need some easier tools to dip your toes into.
If you ever run into any hiccups, pop your head into here, or ask one of our ever-talented guides ❤️
Until then, god speed and may the dice be ever in your favour
Lmao, way to barge into that touching conversation with a LITERAL RIOT OF ENRAGED GOATS ON WELFARE
Yes
I don't know if there's a place to mention this, but I figure this is the most reasonable channel? For the latest Mini-Poll about desired 3.5 improvements, I chose "other" (in addition to two others), but is there a place to specify what other thing you're looking for? Personally, the way Suno often (though not always; but often) tries to fill the full generation time, even if the song should end earlier, is a big annoyance for me that I'd like to see improved. (Yes, I know I can "extend from" or cut the ending in another program; but I'm trying to automate as much as possible, and that's a manual process.)
https://discord.com/channels/1069381916492562582/1252783540131860503 there’s a known-suggestion for that very idea, if you wanted to throw an upvote onto the post in the link? That way the devs know it’s still something people want — out of ~400k, hard to believe only ~70 have stopped in to say that’s what they want 👀
I already have, awhile ago. Ah, well, if that's the only outlet to communicate this desire to the devs, so be it. Personally, I don't even need control over the length, I just wish that when it finished singing the lyrics, it was always smart enough to drop an outro and stop there 😂
is there anything that I can do to help with this issue? Or escalate this issue? Is it some sort of shadowban triggering due to some sort of usage? Its def server related, not frontend It is blocking POST requests attached to an account, same browser different emails it will work on one, not work on another.
403 Forbidden requests so far are liking, generating/creating of any kind, however adjusting the playlist are allowed
https://discord.com/channels/1069381916492562582/1252332553738653838
So i have this idea to create a midwest emo song but idk where to begin with it, i was wondering if i can get some tips to help me
for both custom or regular generation, try your best to describe it in concise detail, I like including the BPM and the feel, tone, into the prompting. So for example
A midwest emo song about eating alot of cheese and feeling lazy after eating. 100 BPM, clean distorted chords, technical drums, depressing and grim, guitar solo and hardcore breakdown
you'll get more control with the custom lyrics
adding square brackets for custom is super helpful like
[verse 1]
I ate alot of cheese, and now I feel bad
[chorus]
woe is me, woe is me
[Instrumental Break]
[Bridge]
I hate cheese! I give up!
[Hardcore Instrumental Breakdown]
[chorus]
woe is me, woe is me!
[end]
lol
but generally, square brackets are used for instructing the model to do things to the song, parenthesis is for a emphasis, or backup singer basically (someone in the background)
Alright. thank you!
have fun!
i cant log into my account. I don't want to be charged when i'm not even using it. Can someone help me?
The most common way would be to click "sign in" or " log in" and then "forgot" or reset password to have reset codes sent you. What's the issue out side of resetting the password?
Isnit that you are somehow denied access to your account? I can't see what immediate charges there would be while logged off, other than monthly subscription fees.
i have my user name but i cant find any other login info. I thought i signed in with my gmail account but i don't see any info. I put in my credit card for 10 a month. I can't log in. I wish i could with my username. I canceled my credit card hope that helps
Most sites have the option to remind you of either user name first , then reset password. I'll sign out and then see what my login page looks like in that regard.
On the home page (not signed in) on my android phone, the sign in button is upper right. Then a pop up , I choose to sign in by Apple, Discord, Google or Microsoft, as you had originally associated your phone number and either of those with your account. Optionally the next line down you can just enter your phone number and it sends you a login code by SMS.
i did that but it go's to another account. Is suno a scam. i don't get it
So I only put in my phone number to get a code, to see. (Then just declined and clicked the Google icon to automatically log me in by it identifying my Google account on my Android phone. No where to put in an email.
You did which "that"?
That the log in popup on my android phone, samsung
Which "that" did you do?
yea i logged into with my phone number but it went to another account
You can log out by clicking your profile picture then sign out
If the third party sign in is still logging you into the incorrect account (wuthout asking which to log into), delete all your cookies
i know how to log out. i don't know how to log into my account all i can find is my username
It responds to me "enter the verification number sent to your phone" , that being the phone number you entered. So it sends a verification to another number? Or it's doing something else not expected?
When you signed up for a subscription, you should have gotten an email from Suno sent to your account. Search all your inboxes for "Suno"
If you still cannot find the correct login, you should be able contact either support@suno.com and provide the reference ID for the payment process ID (which you should have gotten in an email) when you purchased the plan as part of the verification, and/or contact billing@suno.com if a suspension needs to be made.
There isn't anything we can do here on Discord to retrieve what account you signed up with
Ahhhh.... are you using the app?
im on pc . web browser
There is a web page that looks like it's suno.
Hold on.
Stupid me. No app. But the page is simply "suno.com" no other embellishments to the name.
If it's a white page, it's not what you're looking for. Suno has a black background, least on my phone it does.
There is home page on mine.
This is a valid way to use Suno. As long as you're paying for a subscription, you own every beat it makes for you so you can get inspiration from or even copy them exactly using your own instruments.
Billing@suno.com is the help email.
thank you
And on Apple IOS, the app is a white background apparently. Click the buyer menu bars top right and click HELP to read and FAQs. But write directly to billing as well to see if they can help further.
Do you use more than one google account in that browser? If so, you should try logging out of all of your google accounts (via Google, not Suno) and then starting at suno.com to log in via google and sign in. It's probably automatically logging you in under your currently active google account.
Ooops... click the *burger menu bar top right... (sorry typo) and it lists HELP, ABOUT, ETC.
i have 3 emails i can't find any info on suno.com in there but i did pay for it about a week ago for 10 dollars a month.
hello
I have the same issue on some websites. Like. OF will just create a new account for any google ID you try to use, so you can't just try one to see if that's the one you originally used. You'll end up with 5 accounts!
So go to google, log out of all your google accounts. Or, alternately, go to suno in a private / incognito tab and log in via your gmail that way.
i have one google account there is no info on suno
What Tinda says. try that as well. But it's odd that by using a phone number , you get a code that leads you to log in to something that is not what you recognize as your account or it doesn't go anywhere.
Did you check your bank account to see to whom /where the charges went? Often those would list the web page or phone number.
Google wants to be helpful so if you're already logged into an account you've authorized with Suno, it'll just use that one rather than checking to make sure it's the one you want.
You said you have 3 emails but only one google account, is this correct?
i have a hotmail, a gmail and yahoo mail
The subject line of the email receipt you get after subscribing is Your receipt from Suno #0000-0000 (where the zeroes are just your invoice number).
Oh. Wait....!!!!
👋
While we're on the subject: y'all need to install BitWarden.
I just tried to sign back in and it through me into someone else account. Uh oh.
with your phone number?
So it may simply be a glitch, Defsleep
can you access subscription info fo rtheir account, etc?
Yes. With my phone number. I got an SMS.... PUT THE CODE IN.... and it appeared to do nothing.
Yes. I am in their account. So I'll log out and restart the phone.
And you previously logged in with your phone number?
Oh yea? True. 50 credits. That may be a new account. Odd.
It's doing what I said before: creating a new account for your phone number.
@polar spade if using a browser, might clear cache?
yea all i got is my username. no idea how to log in
Well, the issue here isn't cache-related. It's that using a phone number created a new account for someone who previously logged in using only their phone number, if I'm getting the full context.
phone number does not work that just logged into a different account
Ah ha. Right. I may not have used that number previously, so it simply set up a new account. That's silly. Ok. I'll log out and click Google instead.
And you searched your emails for the subject line I gave you?
@mystic flume copy
Yesm, sorry
Yay ur fixed!
Now install bitwarden.
Dang all these people I see have thousands of plays and hundred of thousands likes
Back to Defsleep.....
Ignore that; they're just playing for views. It's weird.
Mine are like 3 likes! 60 plays!
have a good night hopfully canceling my credit card will fix it
Playing for views? Ya I'll just ignore it. :p
Well, that'll stop payments yes. But it won't fix anything.
Sure. Go in the showcase channels and watch. People just posting their music over and over begging for likes.
great i'm starting to hate the site
I've offered several solutions but I can't tell if you've tried them or not.
goes and looks for showcase
i tried everything thanks again
Well, as long as the % is between .5 and 2%, then that's what I see the "trending" scores. But so many likes within so much time is more likely the benchmark.
% meaning, likes / plays
How does someone add their song to showcase?
Meh, it's a fool's errand trying to game that system. The only people who benefit from it are Suno.
you just paste the link. But I'm telling you, it's a waste of time unless you wanna do what they're doing, and get bots to like your music on the site.
Oh, ya no, nvm!
Lol thank you
I don't see a benefit for participating, personally.
@mystic flume you work for Suno?
Or just help em out?
Instead I focus on writing good songs and then making them sound as good as they possibly can, and then release them out into the world to make their fortune.
Neither.
Your % is .05% , so you are trending. Yay! But if no more likes, and no more plays in so much time passed, that trending might drop off to limbo again. Just a guess of mine, why some jump to trending then disappear.
Oh, so you just help people for fun? 😉
I'm just a girl standing in front of a boy, asking him to not participate in the weird popularity contest that has no prize.
I'm just a boy sitting in front of a girl not looking for a prize! :p
Its purpose is to draw people to the service, so I get why it exists. I just don't see how it's beneficial to participate when I could be doing so many other things.
I threw a few in public just for fun, but will likely remove again. Same here, I'm using to pitch to artists. So there is no benefit to be heard here. Keep it private and professional.
But I do a lot of post work / editing / mixing on all my songs so I don't even really like the original Suno versions except for a few that somehow emerged fully formed.
I just create songs. Lots of em.
I have ADHD and can't sit long enough on a PC to create music, or play an instrument.
I can how ever write. Dunno why.
So I write music to put in Suno.
Suno is amazing.
It is amazing, yes. You write your own lyrics?
70-90% yes
You write "music"? And upload the audio?
I don't upload audio, nvr tried it, I let Suno do that
I just write lyrics
I think my voice would only scare people
Oh you said you write music? Did you mean lyrics?
Yes sorry
Lyrics... got it. Ok. Right. Yes. Cool to see what it does to lyrics.
Facts!
facts?
"cool to see what it does to lyrics"
Yes, facts.
This isa first. All our avatars look.like actual people. Lol
Ahhh... word. Got it.
Lol, mines a pic of me in a zombie apocalypse
Don't let mine fool you I'm a very sophisticated AI from the future sent to make great glam-rock.
I'm just me. Not super intelligent, but smart when it comes to common sense, and a jack of all trades kinda guy
Future Droid writing past genre. Cool.
<<< presently writing americanized kpop songs. Who knew. But suno definitely can help the production and give direction on the melody. So it's something.
I like it!
I make random genre, all kinds.
For many people! Love keeping there chins up!
Making them happy when they are low, etc!
I am one of them glass is half full kinda guys though
There ya go. Good attitude. Well done.
Thanks! You too!
Even half empty, it means it was once full, so someone benefits either way. Lol
Agreed.
.... as is my brain.
Lol reguardless! There's water! "Drinks it" hahaha!
Well this is tech-duscussion. You should pop over to the general discussion and get some opinions on yours creations.
*your
how do you get the songs to be less long, every song is going almost 4 munutes and they often change style in a way that is very annoying at around 2:30
Agreed, it just kept telling me to come here for some reason.
@turbid lintel chance.
😦
Would be cool if you could put a time limit!
They are aware that we often want the sing to end somewhere soon after the lyrics are done. Using commands like [end], [ Fade out] , or [outro] might help.
the hard cap at 2 minutes was even better
@polar spade there are commands!? Where do you find this info!!??
you just make them up
here is my latest song, it goes really good until they just add on a total different style around 2 minutes at the end you can see the random stuff i put in the lyrics https://suno.com/song/7935670f-a757-4a0c-88b4-6ec7ed98c404
experimental abstract African folk deconstruction noise, lo fi, drone in background, vocals low in mix, twelve tone song. Listen and make your own with Suno.
If you extend a song, it then caps at two minutes for an rural, song with words, song. You can try a hack by working out how long the song should be, and extend it 2 minutes from your songs proposed ending point. It's odd, but could work.
trying to get good random distortion
Since it does often repeat 2nd verses, and makes up it's own machups of lines, it can make the song verbally longer than you wish as well. Sometimes good, sometimes not so much.
Have a good evening.
Try extending from the beginning of an instrumental you like, then. You'll get the benefit of v3.5 with the 2 minute cap you want.
You aswell.
The most effective way to get your songs to end when you want is to write the lyrics in a cohesive structure and then modify that to augment Suno's original interpretation once you find an interpretation you like. Don't try to get a one-gen wonder.
*rural=aural
How do I end a song in Suno 3.5?
Extend where you want to end with the new extension set to either v2.0 or v3.0 when using ending tags (e.g. [end] in lyric box and "end, short" as the style tag) as they are much more reliable for ending than 3.5.
(Short to help end it quickly in hope for credit refund)
Yes. Exactly. I write in the style of the genre I'm trying to force it to recognize by the writing style. Then prompt in that direction as well. It's usually on the mark, but unfortunately still like to repeat 2nd verses or double up the choruses. So then it's extension time and 19 gens later, ta da!
I think of it as a collaboration. I write lyrics, then Suno writes a song from then. Then I have the power to change it a bit at a time by adapting to the generated structure. It's like a railroad. Suno finds the easiest way to navigate the obstacles between the song's beginning and end to achieve a sound it thinks is "good" -- that's a strong limit on how much you can change things. Trying to make one change may be like trying to turn the railroad to go straight through a mountain. Other times, you can find a nice flat spot that allows you to add a pivot naturally.
Very well said.
The initial generation is like the terrain you have to traverse. So, if there's a river going through it and your plans (lyrics) don't account for a bridge to go over it, Suno will create a bridge using what it has.
So far, I'm amazed at many of the results simply as they are. I crashed my laptop, so when I get a new one and get my daw back up, then I can do more to clean up the gens, same as you.
This is not scientific (or at all accurate) but it's a simple model that helps me understand how to manipulate Suno to do what I want.
And yeah, the results amaze me too. But it's a long way from "amazed" to "a song I want to listen to on repeat"
Technically, I've now found it simply will repeat old versions of lines, regarless of my changing the lyrics. So i now clear the "create" window every chance I get before posting extentions and it really helps to keep it updating properly.
Speaking of repeat. They removed the repeat button. I'm sad. As a practice, I listen to songs at least 50 times on repeat to get so sick of the bad spots that clearly need to be re-written. They become evident the more you listen. So, from my perspective, it's made using suno a bit more frustrating.
Every song I've written gets somewhere over 10+ rewrites. It's just naturally built in to the process.
I do the same. Well, not using Suno's player, but the concept. The first "final" version is usually the one I'll use to make sure I like the lyrics and all the words are pronounced correctly, etc. Then I go in and do a bunch of extends from various parts of the Full version, and edit those into the master in Audition.
Test #34: if you can't sing the song you've written and can breathe at the same time, though a novelty and fun to listen to once or twice, it's not a good song for others to listen to.
Once a well-written song, you throw it into many other genres and get another good song out of it. Save the 7-syllable, 4 line stanzas that can be played in any genre of course.
Good to read someone who uses the site for serious endeavors. Nice to meet you. I'm off get some sleep. Good night.
Support@suno.com if you need someone to look into the matter for you — sorry we can’t be of more immediate assistance
You don’t “command/code” Suno, you “talk to it”.
https://www.suno.wiki can show you [metatags] to use, but they are just “suggestions” that Suno can easily ignore if you give too many
Ty
You are very welcome! #1252267444773326848 is a very helpful wealth of information, many users share their observations
Feel free to ask myself or the guides any questions — we’re happy to help you learn and experiment with Suno 
Love you guys
Also, only the people with purple names are staff — everyone else just loves Suno.
It “woke me up” from “not feeling like myself” for a long time — I want to help even one person do the same.
Back to bed for me, but I wanted to introduce myself, and I’m glad you got to meet my rockstar of a friend Tinda Lee — she sure as heck knows her stuff 👀
I'm colorblind, but thanks, have a good night! Thank you!
IMHO Every one can learn to sing if his vocal cords are all right.
Lol thank you, maybe some day I might try.
That's cool. Most of my songs are not in English, so it is harder to create rhymes. AI has really problems to create rhymes in my language (Slovan).
I have created interesting house song with perkuse I'd like to publish this but is there some free online service to master the noise and make the tics sound better. I cannot describe the sounds which I need to emphasize from the percuse, its like cans and hihat (closed). But there is too noise, and the cans and hihats is too indistinct. I am looking for servise which will not require my credit card on sign up. Just free.
Good point please send this to suggestion...
why dont you just use the custom mode for specifically "jazz" style? 
the Generate Lyrics box does exactly the same as auto-generating lyrics from simple mode does!
the nihongo song is missing correct accent in japanese.
I've yet to see Suno respond to tagged instructions like that in a predictable/repeatable manner. I do think that Suno tries to do something with some of it; just usually not what we're expecting. Adding symbols/characters to the mix is much the same. In my mind, it's all just Prompt Noise. Having said that, I use Prompt Noise all the time... it can result is some interesting and wild tracks.
Here's a remix https://suno.com/song/6f393b1a-0890-4e67-9905-fa65f14caf6a
experimental abstract African folk deconstruction noise, lo fi, drone in background, twelve tone song. Listen and make your own with Suno.
To further that point, it’s made even more difficult because the “command prompts” often work as “thematic prompt decorations”
Like saying “<[bpm:200]>” can suggest that the song should be fast or speed up, but it doesn’t act as a dial so we can necessarily choose the bpm.
You use prompt noise / prompt decoration pretty frequently, both in the [metatags] and style prompts, so I value your opinion on the matter — similarly to Stable Diffusion, we’ve done a lot of experimentation seeing how a single prompt decoration can “randomize” or influence a generation based on how thematically weighted it is.
One other thing that I've noticed, but haven't specifically tested, is that it ***seems *** like the more conversational or potentially lyrical the descriptors are within tags, the more likely it is that Suno will break out in to song with them, especially on the longer generations.
That has been a thing since v2, but it's easy enough to get carried away with tags/styles, and I believe(?) that I've seen instances where shorter style-prompts have been read out as lyrics.
I had some interesting results when I removed the lyrical noise block, leaving Suno with nothing but the tags. This one is close to what it sounds like in the room when I'm watching the television with headphones on and a guitar on my lap https://suno.com/song/8b9f4394-eed9-41ae-99c9-c26833faf4f3 - even that opening drum hit could be from the tv room, as there are bongos within arm's reach 
experimental abstract African folk deconstruction noise, lo fi, drone in background, vocals low in mix, twelve tone song. Listen and make your own with Suno.
Yeah, I've had that happen a lot with "Aggressive" when there are only one or two other descriptors
perfect instrumental f'd up by some dude shouting AGGRESSIVE!!!
You're just intentionally trying to confuse Suno with these metatags now -- watch your utter horror unfold when v6 actually nails all of these requests somehow
hahah