Uberduck Written Tutorial
  • ‼️BEFORE YOU ASK ANYTHING, READ THIS FIRST!
  • 😊Introduction
  • 🗒️Plans and Updates
  • Discord Server Usage
    • Uberduck's Discord server walkthrough 100%
  • Requirements
    • 🤔Which Sources Should I Start?
    • 🛠️Requirements
  • Gathering Dialogs
    • Intro
      • Audacity Tutorial (WIN/MAC/Linux Users)
      • RX7/8 Tutorial (WIN/MAC Users)
      • FL Studio Tutorial (WIN/MAC Users) (WIP)
      • Vegas Pro Tutorial (Windows Only)
      • FFmpeg/FFBatch Tutorial (WIN/MAC/Linux Only)
      • Web version (Any platform)
  • Transcribing and fixing audios
    • Transcribing Dialogues
    • Transcribing Dialogues (Web Version)
    • Templates
    • Vocal Remover Colab Page Guide
  • Dataset sharing
    • Compressing Dataset
  • 🌮Installing Tacotron 2 Locally
    • 🐍Python Clean Install
    • 💻Miniconda Setup
  • Training time
    • Easy or Hard?
    • Training Audio With Tacotron 2 Colab Page
      • Simplified Training Page
      • Advanced Training Page
    • Setting Up for Local Training (WIP)
  • Testing Audio model
    • Testing Voice Model
    • Testing Voice Model (CPU)
  • Troubleshoot and bugs
    • Troubleshooting
  • Extra Stuff
    • Talknet usage
  • Wrapping up
    • The end
Powered by GitBook
On this page
  • First things first.
  • Finish

Was this helpful?

  1. Transcribing and fixing audios

Vocal Remover Colab Page Guide

PreviousTemplatesNextCompressing Dataset

Last updated 3 years ago

Was this helpful?

Sometimes audios may have noises and background music (which let's say your audio is 2.0 and not 5.1 which I talked about on-page of "")

However, we can use a thing called Vocal Remover. Basically how this works is that AI can try to remove noises at it can so it can grab voice and instrumental to it. Aka, well you know. Vocal DIY isolation.

First things first.

The first thing we wanna go is this:

This will redirect you to the vocal remover colab page. Once you are in there you will be greeted by this page:

So basically that's the main page, what you wanna do next is to click copy to drive, so that it'll save your settings in your drive.

Next up is "Mount to drive" cell. If that checked then you gonna have to log in to your google account, don't worry about it, you know list things and stuff it's google after all... copy and paste code to it and it will install stuff on your drive. Otherwise uncheck and it'll install some stuff on google colab.

What you wanna do next is to upload your files or one file into a specific folder. Folder by the name tracks. You put it in there and you should be good to go. On google drive it should be a folder like this:

Otherwise if not then go inside the vocal remover folder and select tracks. Then upload in there and it should be good to go.

Once that's done select press me cell, this will load lists of things of what it has:

The file name is what we need.

So now scroll down until you find this section:

What you wanna do is:

  • Replace input into your audiofile name that you saw in press me cell section. Or use youtube url for it.

  • If you have that many audio files then check convertall

  • Keep the pretrained_model as it is or you can change the model from list if you didn't liked the result.

  • Window size 320

  • keep parameter as it is (parameters are also in the list if you didn't liked results)

  • keep high end process as it is

  • Aggressiveness 0.09 (you can experiment with it if you want but don't put it too much)

  • Checkmark the tta

  • Checkmark download is optional so that it could automatically download for you.

Once that's done you can now click play on this cell and wait until it does magic.

And once it's finished you gonna have to go to seperated folder:

Vocals are the only thing you are looking for but you can also download instruments too.

Finish

If you've done everything correctly then great job! Now for the final part, training time...

Training time
https://colab.research.google.com/drive/1eK4h-13SmbjwYPecW2-PdMoEbJcpqzDt?usp=sharing
which sources should i start?