What's new
By:
Filters
LiteRECORDS

Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • Guest, before your account can be reviewed you must click the activation link sent to your email account. Please ensure you check your junk folders.
    If you do not see the link after 24 hours please open a support ticket.

Powerful New Vocal Remover AI - Instructions

Updated the guide to follow a new method for converting tracks in the cloud. Tested several times and works as expected.

This should work if you have issues running it locally and works on all operating systems.
 
the cloud thing is dope! i got everything done except step six... what am i supposed to do there? what is this workflow supposed to look like? right now i upload to google drive, edit the audio file portion to match in the fifth step

#FIFTH STEP

!python inference.py --input "/content/drive/My Drive/blah blah blah.mp3"

and then after about ten minutes the output files are found in the vocal-remover folder. then i download and then go and delete the original and output files. is that how it should work?
 
Last edited:
dimkeal said:
i got everything done except step six... what am i supposed to do there?
Click to expand...

For each step just click the little play icon (Run cell). The sixth one is to export the tracks, otherwise without it, you wouldn't find them anywhere saved.
 
so everytime i want to convert a track i open the collab and my drive and edit step 5 and run cell to convert? just checking if that is the workflow.
 
dimkeal said:
so everytime i want to convert a track i open the collab and my drive and edit step 5 and run cell to convert? just checking if that is the workflow.
Click to expand...

If you already converted one track and want to do another, just edit the fifth step (the name) and proceed with the rest. If you haven't used the tool in a while and you already closed the tab with the Colab page, you would need to run all of the steps again, but I don't think they take that much time to complete. 1-2 minutes in my case.

The reason is, you're using hardware that is located in another place and if you don't use it for a while (occasionally you'll be asked if you wish to reconnect so that you don't lose connection) or you exit the webpage, that PC is no longer attached to you and you need to connect again, hence run all the steps.
 
djtayz said:
If you already converted one track and want to do another, just edit the fifth step (the name) and proceed with the rest. If you haven't used the tool in a while and you already closed the tab with the Colab page, you would need to run all of the steps again, but I don't think they take that much time to complete. 1-2 minutes in my case.

The reason is, you're using hardware that is located in another place and if you don't use it for a while (occasionally you'll be asked if you wish to reconnect so that you don't lose connection) or you exit the webpage, that PC is no longer attached to you and you need to connect again, hence run all the steps.
Click to expand...

Where exactly do have this and can you please link to your project? Very curious to check it out.
 
Anjok said:
I think I figured it out! I tested my instructions on a VM running a bare-boned version of Windows 10 and got the same error you did. The issue is some systems don't play well with PyTorch's Cuda edition (which is the one I linked). Also, many people have various encoders already installed that play well with Python and some don't (especially if we're dealing with someone who's installing this on a fresh version of Windows), so I was getting SoundFile errors too. Installing soundfile fixed that issue though! Please remember to install and configure ffmpeg before trying to run mp3, mp4, m4a, and flac files through this tool!

Leave your installation as is and simply install the following and you should be good. Please let me know if this works for you! I also updated the main thread with these instructions. One thing though, don't check the GPU option is you don't have a stand-alone one installed.

pip install torch==1.5.0+cpu torchvision==0.6.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
pip install SoundFile
Click to expand...

That fixed it!!! Hell yeah. Even better, it takes less time with GUI and multi-genre model than command line for me. What would take about 4 to 5 hours with command line method now is less than an hour.
 
NewAgeRipper said:
That fixed it!!! Hell yeah. Even better, it takes less time with GUI and multi-genre model than command line for me. What would take about 4 to 5 hours with command line method now is less than an hour.
Click to expand...

Dude glad you got it working now I'm gonna get my other laptop up and running
 
===============================================

Google Colab Method

If you encounter issues during the installation of the regular vocal remover or the GUI, you can use this method. It doesn't require any special or powerful hardware, as all of the processes are done in the cloud. 1 track gets processed for around 8-10 minutes.

STEPS:


1.Download this archive and upload the folder "vocal-remover" to your Google Drive.
https://mega.nz/file/MewDjTTR#WetXWo7iwI86oj-OMBnE0-ZeF0bmugVjGt1AHggltgs

2.Go to https://colab.research.google.com/ and click on "Upload".

3. Download this file and upload it to Google Colab.
https://mega.nz/file/keghkLTY#enyimMMRNZ1z1fFVLqvGtJJCjZv5ZkZT-0y9x9p3QU4

4. When you open the project, you'll see a couple of steps that I've numbered to be easy to follow. Just click each one and wait until it gets finished.

On the fifth step, where you're choosing which track to process, make sure you've typed its name exactly as it is + the file type (WAV/FLAC/MP3 etc.)


5. After completing the sixth and final step, your instrumental & acapella tracks should be found inside the "vocal-remover" folder.
Click to expand...
Now this is very useful. 8 min 16 sec vs 45 min? Hell yeah [MENTION=26971]djtayz[/MENTION]. Awesome work. Great to have all these options in this.
 
Last edited:
[MENTION=39673]Anjok[/MENTION] [MENTION=26971]djtayz[/MENTION] you guys rock. This is only gonna get better and turn the music community on it's head.
 
Hey guys, i think you should ask if it's OK to post a donations link on this thread. I would donate for sure to help with the cloud space storage.
 
NewAgeRipper said:
[MENTION=39673]Anjok[/MENTION] [MENTION=26971]djtayz[/MENTION] you guys rock. This is only gonna get better and turn the music community on it's head.
Click to expand...

Thank you! I'm very glad it's working now!

I've always wanted a tool like this! I love instrumental versions of songs for a plethora of reasons including making mash-ups, easier note transcription, studying, or just enjoyment. With that being said, it was natural for me to want to share this technology with those who appreciate music as much as I do and make it widely available to the music community. Because if I was on the other end, I would be hoping for someone to do the same if they were aware of the technology like I was. I didn't want a powerful tool like this to be out of reach for anyone. Not to mention 2020 has been a rough year for the world, we all needed this!

I'm glad I found a community here that loves music as much as I do! I hope you all enjoy this AI!

I greatly look forward to the explosion to creativity this will inevitably lead to.

Also, MASSIVE kudos to the GitHub author of this AI!! It's geniuses like that who work instrumentally (no pun intended ;) ) to bring technology like this to life!
 
Last edited:
djtayz said:
Stems? Are you referring to Demucs?
If it's about this one, you need to create your own, the tutorial is about that.
Click to expand...

Ah damn. Was hoping another cloud tool for it was happening. I can't make heads or tails of most github descriptions. But if you can post a tuto in laymen terms I'll be grateful.
 
djtayz said:
I can make a guide on how to install it on Google Colab, since it won't require coding there, otherwise this is the GitHub page and there are a few examples and comparisons https://github.com/facebookresearch/demucs

It's my go-to way for DIY stems. Downside is on some tracks it can ignore background vocals and leave them as synths, rather than part of the vocals track. But depends from song to song.
Click to expand...

This is what I meant by Google Colab. You mentioned earlier you could make a tuto on Colab. One for dummies would be great.
 
You must log in or register to reply here.