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
  • "What if I haven't done with audios"
  • Preparing for transcribing dialogues
  • Creating a text file
  • So what now?
  • Finish

Was this helpful?

  1. Transcribing and fixing audios

Transcribing Dialogues

PreviousWeb version (Any platform)NextTranscribing Dialogues (Web Version)

Last updated 3 years ago

Was this helpful?

So now you have done the audio it's time to transcribe them. I know I know. It may take some time but hear me out for a second. If you won't do transcribing dialogues you'll then get some errors. Especially if you don't specifically write correct dialogue/lines.

"What if I haven't done with audios"

If you haven't done with audios then what are you doing here? Go back in there!

Preparing for transcribing dialogues

So your files should be looks like this:

Once you have done with audios we'll need to create a text document. If you don't make one you'll end up with errors and you won't gonna have training without a text file.

Creating a text file

What you want to do is right-click and select Create, and create a text document:

What you want to do now is that name whatever file you want but as long as it's not long. For example, we will be naming into a list.txt but yours might be different if you want a different name.

And it should be looks like this:

You open this bad boy up and it'll open with a default notepad. But I don't need default notepad so instead, i'll be working on notepad++

So what now?

The next thing you wanna do is to take a look at this, it's empty, right? yeah, that figures. What you wanna do is to do this:

wavs/*filename*.wav|*Dialogue*

Okay, let's start off with something interesting.

First off we have wavs/*filename*.wav. It tells tacotron on where those audio files should be trained (well it won't be waves but you'll see it what i mean) and the *dialogue* obviously you need dialogue, if you won't write one, your model will be gibberish and not understandable so be sure to add dialogues this time. Here's an example for list.txt that I am working for my own model:

And once you have done you'll now have to save. Make sure dialogues is the same as the character would say but also if you have subtitles lying around then this might be much easier than that.

"But what about the characters like ?, !, comma, etc."

Sure, it works like that. Some characters do work but some of them for example spanish, french, and other letter symbols won't work. so you gonna have to workaround.

By the way, every dialogue should have a period on it. Which is DOT.

Finish

If you have done everything right then congratulations, we are now going for the fun part, which is training!

Training time
Gathering Dialogs
Ooooo, that's alot of audio files.
Create, text document. Again my OS is russian :P
Looks empty right now...
That's a lot...