🎙️ ToneSwap: Real-Time Voice Conversion App

ToneSwap is a real-time voice conversion application that transforms a speaker's voice into a target voice while preserving the speech content, tone, and emotion. It supports various audio formats and provides a user-friendly web interface via Gradio.

---

🚀 Features

• 🎤 Converts your voice to another person's voice using voice samples
• 🧠 Zero-shot voice conversion using pretrained models (FreeVC + WavLM + HiFi-GAN)
• 📁 Supports uploading or recording source and target audio in any format
• 🔁 Automatically converts input to mono 16kHz .wav using FFmpeg
• 🌐 Easy-to-use Gradio web interface

---

📦 Prerequisites

Before running the app, download and place the following files and models in the correct directories:

✅ 1. Pretrained Checkpoints

Download FreeVC model checkpoints and place them in:

checkpoints/

📥 Download Checkpoints

---

✅ 2. WavLM-Large Model

Download the WavLM-Large model files from the official Microsoft repository and place them in:

wavlm/

📥 WavLM GitHub

---

✅ 3. HiFi-GAN Vocoder (Optional, for SR training or fine-tuning)

Clone the HiFi-GAN repository and download generator_v1 pretrained model. Place it in:

hifigan/

📥 HiFi-GAN GitHub

---

🧪 Setup Instructions

1. Clone this repo:

   git clone https://github.com/<your-username>/ToneSwap.git
   cd ToneSwap

2. Install Python dependencies:

   pip install -r requirements.txt
   

3. Install FFmpeg (for audio format conversion): Windows: FFmpeg Download
4. Install Gradio:

   pip install gradio
   

---

##✅ How to Run

python app.py

⚠️ Note

Due to GitHub's file size limitations, this repository does not include pretrained models or checkpoints. You must download and place them manually as instructed above.