Strangely, attempting to venv this is giving me fits. Kept claiming modules were missing (despite me doing "pip install .") kept trying to install the missing ones, but still couldn't get it to run. Now looking at the error "from dac.nn.layers import Snake1d ModuleNotFoundError: No module named 'dac.nn' " Happy to do more research and digging myself, but cursory searches aren't turning anything up. Suggestions?
as an indie game developer, this is super useful. even when you have access to thousands of sounds to use, finding exactly the one you want takes forever because you have to listen to every , single, one of them.
Everything is a file searching app that works on windows. You can find anything with it, as well as limit it to search for just or audio. Very useful for that kind of work. But everything has to be named right!
@@Nandarion learn to use Vital Synth , it's free.I dont' find any of this stuff sounds any good. I make my own custom sounds for my own games. there's also a tutorial playlist on how to do this for games using Massive Synth here on youtube.. MAssive synth goes on sale all the time over at Plugin Boutique.. i keep looking into this ai stuff and it's just not useful for what i do.
@@Nandarion You can just add echo to any sound in the game engine, so its the same sound. If you need it to be longer, you can sometimes just copy the end of the sound and paste it onto the end.. and then try to blend it in to the original. I did that with a siren sound recently, it works pretty well. The real tough thing is finding that perfect sound you want, but it has other sounds you don't want, or bad quality. I'm still hoping for an AI to come along that can take a sound and text, and render a new version of it. that would fix the noise problem, and you could make longer versions. I feel like this is getting close.
@@Nandarion to make it longer, just make some duplicates of a middle part in audio editor and then just generate new one using init sound feature and low noise. Echo, reverb etc. is better to pass it to game engine, just a suggestion - keep src sound clean, without any effects.
When I saw this appear in my timeline, I was a bit scared for a bit. Then I just realised I could include it in my workflow since it opens many possibilities. Gotta toy with it someday :)
Thank you this is working... so glad this worked.... I get to where I dread starting new installs/configs etc because I have gone through tutorials that are sometimes over an hour to get through and then they don't work... this one worked on the first try thanks to your detailed instructions.
Hey brother, a new tts model has just came out, it is called "seed tts". It is actually a family of high quality versatile speech generation model. It has some other features too! I strongly want you to make a detail video on it, from showing it's use cases to the actual installation process! It will be very helpful for me! Thank you
First impression: Seems interesting for generating generic sounds, or uncanny sounds. I tried a few: "dog meowing", "people laughing scary", "deep breathing", pretty entertaining in a horror game or something xD Tried other prompts like "dot matrix printing", "dolphin sounds", etc and it wasn't very accurate just printer sounds and sea life sounds. Might have to keep generating to get what you want. So seemed best for generic placeholder SFX or uncanny noises; at least with the default options.
Can you just talk about VRAM or put info in description in your every open source AI video ? This is very important information which is missing in every video you make.
why when I press generate audio it says CUDA not support and the generation with CPU is super super slow. Is there any fix to utilize CUDA for audio generation?
Keep getting Error: could not find a version that satisfies the requirement pedalboard==0.7.4 I noticed that your root folder has the file "build" that was not in the original repository. What is that file, and how did you get it?
Awesome tool, thanks. But when I started the tool first time after installing it, the Win Antivirus alerted a suspicious threat : PUA:Win32/FRProxy \stable-audio-tools\env\Lib\site-packages\gradio\frpc_windows_amd64_v0.2 🤔What could this be? It looks like a French Proxy or something?? Any idea?
Great stuff. Pls could you focus on the settings in your next videos? For example I am trying to figure out how the Init audio works, but no luck. In your other videos you usually cover what different menu and settings mean, but this one seems to be very shallow. Thx!
3090 RTX 8gb but get error cuda memory even at 5 seconds of audio, everything is installed right, how much mem does this actually use? Any tips to get past the cuda memory error?
Please make the model training video. It's providing me with some hit or miss results, but nothing really hits the spot like the preset wavs that I have.
Thanks for making this. Your video makes this tool far more accessible than Stable Audio's readme which suffers from typical geek's blindness to non-coders limitations.
You getting winerror 126? I got stuck there too, people online were saying reinstalling pytorch would fix it.. I tried updating it but still no go on it :/
@@lemda500nm8 if im not mistaken for watching any tutorial. try "pip install ." it mean it will be install anything that program need. but i still got differend error and i gave up.
"pip install safetensors" that works for me later this thing say me something like: "No module named huggingface_hub" And i write: "pip install huggingface_hub" and later some gradio and i install all the thing with "pip install" again and again Now the AI works for me
why does it always generate about 1 min where the actual sound is at the beginning according to our settings ? E.g. 10 seconds of sound , then blank to 1 minute.
i like to think of this channel as an advocate of Accelerationism. he knows fully well this will bring the end of us and thats why he shares it. im all here for it
As said, I can’t really figure out what would be questionable stuff to listen to in a work environment. Visual things are easy, aural not so much. Well… aside from ”that one guy” who keeps playing Simon & Garfunkel on a loop in their headphones … but that’s about it. Truly my lack of imagination.
@@uttula ....I can There's plenty questionable sounds Starting with moans and ending with pretty... unambiguous slapping flesh sounds or slurping sounds... So...em...yeah...
how do you do it? can you message me? and I don't honestly get locking bat files behind a paywall. I could see if it was an exclusive model for members or something like that, but a bat file?
my generated audio is ALWAYS 47 seconds long, no matter what. after the 5 second sound ist played, there is just 42 seconds silence. thats why generation always takes the same amount of time no matter how many seconds i have set
a lot of artifacts and audio degradation,i'm an sfx artist and the quality is not good for video at this moment,need improvment or a new model,there is something good but non at all
Bro you should atleast put the the things we need to copy paste in the description, i know u wanna make sales but you are going way to fast and pasting info you do not provide unless we type it out🤦🏽♂️
So I installed it manually and it works, But how do I run it the next time after i've closed it all down as there is no bat file like the other AI tools have. Thanks.
Wasn’t super impressed with the animation tweening AI in the last video as the interframes still looked pretty bad quality wise, so I think Animators are still safe (for now…). THIS though…. Yeah, I’d say the stock sounds industry is in trouble! 😳
They haven't added the sketch stuff yet, animators will be fine but the labor of coloring each frame is going to be reduced. Which means more anime and some animators will not have to sleep at the studio to get the anime out on time.
I was able to make it work, but couldn't get it to work with cuda. It seems flash-attn is not installing properly for me. Using CPU is just not practical. Edit: I installed on a PC with 4080 and it's much faster. I guess 2080 wasn't supported. It generates using GPU now. Still no luck with installing flash-attn so I dont know how important that is.
@@justinbishop9584 I installed on a PC with 4080 and it's much faster. I guess 2080 wasn't supported. It generates using GPU now. Still no luck with installing flash-attn so I dont know how important that is.
I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
@@danieladler3210 I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
@@justinbishop9584 I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
"pip install safetensors" that works for me later this thing say me something like: "No module named huggingface_hub" And i write: "pip install huggingface_hub" and later some gradio and i install all the thing with "pip install" again and again Now the AI works for me
What's "uncensored" and "NSFW" about a sound effect generator that doesn't do voice, as you specifically point out that's a different AI? Without vocalizing any actual words, stubbing your toe or suffering through passing a kidney stone is indiscernible from moaning for a different reason. Nothing NSFW about it.
salut confrère francophone je vouslais te dire que tes vidéos sont cool et bien faites. Et que ca me fait rire que a chaque vidéo tu mette dans la minia "NSFW" pour attirer du monde 😆😆
0:25 Nooo! It's not from scratch. You do know that this AI stuff uses a MODEL right? A MODEL that is created BASED ON A LOT OF DATA. I'm so frustrated when people don't understand how things work.
HELLO HUMANS! Thank you for watching & do NOT forget to LIKE and SUBSCRIBE For More Ai Updates. Thx
Strangely, attempting to venv this is giving me fits. Kept claiming modules were missing (despite me doing "pip install .") kept trying to install the missing ones, but still couldn't get it to run. Now looking at the error
"from dac.nn.layers import Snake1d
ModuleNotFoundError: No module named 'dac.nn' "
Happy to do more research and digging myself, but cursory searches aren't turning anything up. Suggestions?
I want to extend a sound according to a prompt. Similar to what Udio does, but with my sound not theirs.
hentai sounds ... UwU ♥
Consider pinning/highlighting the error comment thread from "@cassianopaulo1", they helped some folks troubleshoot manual fixes.
can you make a video on how to make google collab pods so we can set up such things ourselves?
as an indie game developer, this is super useful. even when you have access to thousands of sounds to use, finding exactly the one you want takes forever because you have to listen to every , single, one of them.
Everything is a file searching app that works on windows. You can find anything with it, as well as limit it to search for just or audio. Very useful for that kind of work. But everything has to be named right!
@@SwabcraftCreatesthose sounds arent local(most of the time) and need to be loaded online.
@@Nandarion learn to use Vital Synth , it's free.I dont' find any of this stuff sounds any good. I make my own custom sounds for my own games. there's also a tutorial playlist on how to do this for games using Massive Synth here on youtube.. MAssive synth goes on sale all the time over at Plugin Boutique.. i keep looking into this ai stuff and it's just not useful for what i do.
@@Nandarion You can just add echo to any sound in the game engine, so its the same sound. If you need it to be longer, you can sometimes just copy the end of the sound and paste it onto the end.. and then try to blend it in to the original. I did that with a siren sound recently, it works pretty well. The real tough thing is finding that perfect sound you want, but it has other sounds you don't want, or bad quality. I'm still hoping for an AI to come along that can take a sound and text, and render a new version of it. that would fix the noise problem, and you could make longer versions. I feel like this is getting close.
@@Nandarion to make it longer, just make some duplicates of a middle part in audio editor and then just generate new one using init sound feature and low noise. Echo, reverb etc. is better to pass it to game engine, just a suggestion - keep src sound clean, without any effects.
When I saw this appear in my timeline, I was a bit scared for a bit.
Then I just realised I could include it in my workflow since it opens many possibilities.
Gotta toy with it someday :)
bro literally killing every industries💀
💀
sound effects? I can make fart noises
@@zakaris7259 prove it
not him though 🤣it's just the open source community lol
"Bird is chirping while making soft, moaning sounds" .. OK
stable audio training video would be amazing
man this is all im waiting for...
YEAH! i wanna train my own sound fx letsgo!
Thank you this is working... so glad this worked.... I get to where I dread starting new installs/configs etc because I have gone through tutorials that are sometimes over an hour to get through and then they don't work... this one worked on the first try thanks to your detailed instructions.
Hey brother, a new tts model has just came out, it is called "seed tts". It is actually a family of high quality versatile speech generation model. It has some other features too! I strongly want you to make a detail video on it, from showing it's use cases to the actual installation process! It will be very helpful for me! Thank you
Are you sure they released the model
Just what I have been looking for. Thank you AI Overloard!
First impression: Seems interesting for generating generic sounds, or uncanny sounds. I tried a few: "dog meowing", "people laughing scary", "deep breathing", pretty entertaining in a horror game or something xD Tried other prompts like "dot matrix printing", "dolphin sounds", etc and it wasn't very accurate just printer sounds and sea life sounds. Might have to keep generating to get what you want.
So seemed best for generic placeholder SFX or uncanny noises; at least with the default options.
Can you just talk about VRAM or put info in description in your every open source AI video ? This is very important information which is missing in every video you make.
It uses about 14gb of VRAM in my and other people's tests.
@@FenrirRobu thank you
running successfully on a 3060 ti
why when I press generate audio it says CUDA not support and the generation with CPU is super super slow. Is there any fix to utilize CUDA for audio generation?
same, plz
can we get the pip install torch command in the description, finding it hard to read in the video
Keep getting Error: could not find a version that satisfies the requirement pedalboard==0.7.4 I noticed that your root folder has the file "build" that was not in the original repository. What is that file, and how did you get it?
I get the same error
me 2
@@youngcampoproductions make it use a -venv a virtual environment. That might fix it
What a time to be alive :)
2 minute paper
Best Patreon I ever became a member of! Keep up the good work! 👍👊👏
this is great! gonna find out how to train it on libraries I already have for specificity.
Awesome tool, thanks.
But when I started the tool first time after installing it, the Win Antivirus alerted a suspicious threat :
PUA:Win32/FRProxy
\stable-audio-tools\env\Lib\site-packages\gradio\frpc_windows_amd64_v0.2
🤔What could this be? It looks like a French Proxy or something?? Any idea?
Yes. training video sounds awesome! Great job
Great stuff. Pls could you focus on the settings in your next videos? For example I am trying to figure out how the Init audio works, but no luck. In your other videos you usually cover what different menu and settings mean, but this one seems to be very shallow. Thx!
Thank you. Please upload a video explaining your training methods! That would be very helpful.
Well Yes our AI overlords a way to train our own AI Sound generators would be greatly appreciated.
absolutely would love to see a tutorial on training an audio model and also one for running this on comfyui!
3090 RTX 8gb but get error cuda memory even at 5 seconds of audio, everything is installed right, how much mem does this actually use? Any tips to get past the cuda memory error?
I have a question. What are the minimum requirements to run this on local?
I am currently running this on a 3060ti around 60 secs wait for 5 secs generation
omg ty!! this is awesome possum
@Aitrepreneur when do you think/expect we'll have a suno / udio musical local models? Thanks ❤
Would be awesome if there was a tutorial on how to generate you're own model!!
Please make the model training video. It's providing me with some hit or miss results, but nothing really hits the spot like the preset wavs that I have.
I don't know if it's worth it to fight with flash attention package. My 12 GB of VRAM is little to low, and when normal RAM is used things go slow.
Thanks for making this. Your video makes this tool far more accessible than Stable Audio's readme which suffers from typical geek's blindness to non-coders limitations.
ANOTHER HUGE ONE FOR THE TEAM
Great ! And yes, how to train the model video would be Great² 😘
how much VRAM did you need to be able to run it locally? approximately
im running on rtx 3060 12gb pc get a little bit slow, but works fine
@@KimiMorgam I have a 3080 10gb maybe it's not enough 🤔
Well done! But how to run again once it's closed ?
Redo the two specific steps in the video to run it again, the activate environment command 01:50 and the run command 02:40
I can't install it, I get a big error saying it can't determine archive format
There is an official ComfyUI support for this model.
I keep getting the error
ERROR: No matching distribution found for pedalboard==0.7.4
I am also getting this error.
I fixed that error with "pip install pedalboard==0.8.2" but now I get an error when starting it saying no module named 'safetensors'
Which python version you are using?
Some packages need a minimum python version, from the repo it looks like they are developing with python 3.8.10
@@LuizOtavio-uz7jr I'm using 3.10.6, so wouldn't think that's my issue
@@LuizOtavio-uz7jr python 3.12
i dont know why but im stuck at ckpt part..Giving me error every time..Can anyone help?
You getting winerror 126? I got stuck there too, people online were saying reinstalling pytorch would fix it.. I tried updating it but still no go on it :/
Does it work on linux?
Thank you, is very useful
Can video game developers use the generated sounds for their video games that they want to sell later?
Yes, good luck to anyone proving that it was AI generated.
@@TheH1st0ry What do you mean there's plenty of tools that can detect AI generated stuff.
Is there a minimum amount of vram needed? 👀
I haven't tried running this yet, but the model file is only 5GB, so it should be fine.
ModuleNotFoundError: No module named 'safetensors'
What should i do?
getting same error, have you found the solution yet?
@@lemda500nm8 if im not mistaken for watching any tutorial. try "pip install ." it mean it will be install anything that program need. but i still got differend error and i gave up.
well to me it sounds like you put a space somewhere after the period of model.safetensors.
"pip install safetensors" that works for me
later this thing say me something like: "No module named huggingface_hub"
And i write: "pip install huggingface_hub" and later some gradio and i install all the thing with "pip install" again and again
Now the AI works for me
why does it always generate about 1 min where the actual sound is at the beginning according to our settings ? E.g. 10 seconds of sound , then blank to 1 minute.
I've noticed that also. it will be 10 seconds of sound then the rest is all silent
i like to think of this channel as an advocate of Accelerationism. he knows fully well this will bring the end of us and thats why he shares it. im all here for it
Hello future internet, hello AI, hello. This is the start of all of this. *winks*
Every end is a new beginning
What do you mean by UNCENSORED SOUND? is moaning sound or sex sound ?
and the minimum req are?
What a wonderful day
How do I reopen the program after installing and closing it?
reactivate the venv and run the script
I may have absolutely no imagination what so ever, but … what a heck NSFW sound even is?
Not Safe For Work, meaning some questionable stuff... He didn't mention "The Hub" out of nothing.
prob the iconic drum loop intro that plays on every vid
As said, I can’t really figure out what would be questionable stuff to listen to in a work environment. Visual things are easy, aural not so much. Well… aside from ”that one guy” who keeps playing Simon & Garfunkel on a loop in their headphones … but that’s about it. Truly my lack of imagination.
@@uttula ....I can
There's plenty questionable sounds
Starting with moans and ending with pretty... unambiguous slapping flesh sounds or slurping sounds...
So...em...yeah...
Wow you really deleted my comment that told people how to make the bat file to launch it?
he deletes mine too, he want us join his patroen. he became a stupid selfish wanting money
Damn
how do you do it? can you message me? and I don't honestly get locking bat files behind a paywall. I could see if it was an exclusive model for members or something like that, but a bat file?
hey sorry can you help me with install
Wonderfull .. thank U !!!
Can this be used to generate gasps, moans, groans consistent with a given voice?
ABSOLUTELY NOT, but the result is very very funny
my generated audio is ALWAYS 47 seconds long, no matter what. after the 5 second sound ist played, there is just 42 seconds silence. thats why generation always takes the same amount of time no matter how many seconds i have set
Is this Marc Lou?
Please video aout training! Thanks for this technology
The installation worked just fine, but the generation takes over five minutes with my rtx3070 and my whole pc is stuttering. Is this normal?
this worked for me
make sure that CUDA is installed in the venv
it will awesome you provides cmnds in descriotion
Please do a video on Stable Diffusion 3 soon
Can you upload your own samples?
a lot of artifacts and audio degradation,i'm an sfx artist and the quality is not good for video at this moment,need improvment or a new model,there is something good but non at all
You have to use low CFG, High steps, inpainting, and train your own checkpoints. Same as Stable Diffusion
How much vram minimun is required??
24
@@8wedSeriously? Looks like a hoax. It's audio.
@@happy-gq2kw If you only do generative stuff, it takes 8. If you inpaint, it takes 12
i installed it with his bat, but the folder is empty now. I wonder where it is installed now ... lol.
thanks so much for this I was struggling on installing it myself
how to lunch second time for manual users?
Can it make Gregorian chants?
use suno for that
Just tested it and it actually does. It was some pretty scary shit though :)
it works with AMD ?
im new to python how i can install using embedded python
hey a traing tutorial would be very cool!
what are u gonna train first :)))
Bro you should atleast put the the things we need to copy paste in the description, i know u wanna make sales but you are going way to fast and pasting info you do not provide unless we type it out🤦🏽♂️
So I installed it manually and it works, But how do I run it the next time after i've closed it all down as there is no bat file like the other AI tools have. Thanks.
You need to reactivate your python venv then run.
Your content is top notch. Can you explain how you create these beautiful covers for your channel?
Does anyone have any good negative prompts to share?
It installs a PUA reverse proxy written by a Chinese programmer. wll good luck with that lol.
Really? You got the link to the code that does that? I'm curious.
Wasn’t super impressed with the animation tweening AI in the last video as the interframes still looked pretty bad quality wise, so I think Animators are still safe (for now…). THIS though…. Yeah, I’d say the stock sounds industry is in trouble! 😳
They haven't added the sketch stuff yet, animators will be fine but the labor of coloring each frame is going to be reduced. Which means more anime and some animators will not have to sleep at the studio to get the anime out on time.
Uncensored audio? What fuck this even means?
Why is AI audio generation so slow compared to image gen? I thought audio files take up way less space and data.
patreon following because you are the best no joke! also help , i cant create auto run command
HF Space / G Colab?
Custom dataset plz!
I was able to make it work, but couldn't get it to work with cuda. It seems flash-attn is not installing properly for me. Using CPU is just not practical.
Edit: I installed on a PC with 4080 and it's much faster. I guess 2080 wasn't supported. It generates using GPU now.
Still no luck with installing flash-attn so I dont know how important that is.
any solution?
@@justinbishop9584 I installed on a PC with 4080 and it's much faster. I guess 2080 wasn't supported. It generates using GPU now.
Still no luck with installing flash-attn so I dont know how important that is.
I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
@@danieladler3210 I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
@@justinbishop9584 I have the same problem with my 3060. In my case I reinstall drivers for cuda using search in google "Guide To Install CUDA for GPU rtx 3060" there was first link the site justlearnai. I did what was written in this guide (including reinstall PyTorch). Then reinstall Stable Audio. And then problem solved.
ModuleNotFoundError: No module named 'safetensors'
pip install safetensors?
@@baronvonbeandip I have the same problem as the above person. Can you help me? It keeps saying "No module named 'safetensors' "
@@samipya0601 I literally just told you what to try lol
"pip install safetensors" that works for me
later this thing say me something like: "No module named huggingface_hub"
And i write: "pip install huggingface_hub" and later some gradio and i install all the thing with "pip install" again and again
Now the AI works for me
This works to create instrumental songs?
You can simulate these things but making a whole music won't
Alr I'm working at a bar now fuck this
What's "uncensored" and "NSFW" about a sound effect generator that doesn't do voice, as you specifically point out that's a different AI?
Without vocalizing any actual words, stubbing your toe or suffering through passing a kidney stone is indiscernible from moaning for a different reason. Nothing NSFW about it.
salut confrère francophone je vouslais te dire que tes vidéos sont cool et bien faites. Et que ca me fait rire que a chaque vidéo tu mette dans la minia "NSFW" pour attirer du monde 😆😆
👍😀👍
i go tlike 20 errors, bruh
Bro i missed your speeches.
Your a noble alien a greenskin to your blood.
+10000🛡🛡🛡🛡🛡
0:25 Nooo! It's not from scratch. You do know that this AI stuff uses a MODEL right? A MODEL that is created BASED ON A LOT OF DATA.
I'm so frustrated when people don't understand how things work.
Without ChatGPT I would not have been able to install this crap
Could you please talk faster in videos, so boring to wait for instructions ; )
Lol
Comfyui nodes are available 😊
Why TF are you still using windows.... 🤣🤨🙄
Them at M$ are going to find all the potatoes..
Say goodbye to having talent and a soul.
i subscribed just because you are funny as fuck
I thought there's anime moan sound
there... is?
FU
no FU