Voice Conversion Project

Overview

This project is dedicated to developing advanced voice conversion technology using deep learning techniques. Our aim is to transform diverse voice inputs into a single, consistent target voice while preserving the original speech's nuances.

Getting Started

Prerequisites

• Google Colab account
• Access to Google Drive

Installation and Usage

1. Google Colab/Drive Setup: This project is implemented and executed using Google Colab and Google Drive. Ensure you have access to these services.

2. Accessing the Notebooks:

   • main.ipynb is the primary notebook for the project, featuring the successful implementation of our voice conversion model.
   • failed_example.ipynb provides insights into our initial approach, which was later revised due to its limitations.

3. Data and Pre-trained Models:

   • Links to download necessary data and our pre-trained models are included within main.ipynb. Follow the instructions in the notebook for setup.
   • You also have the option to download data directly from YouTube for training. Instructions for this process are provided in the notebook.

Project Structure

• main.ipynb: The main Jupyter notebook with the implementation of the voice conversion model.
• failed_example.ipynb: A Jupyter notebook documenting our initial, unsuccessful approach.

Usage

Recommended

To use the voice conversion process, follow the step-by-step instructions in main.ipynb. The notebook guides you through data loading, model execution, and optionally downloading and using data from YouTube.