AI Voice Replication

A Flask-based web application for realistic voice cloning using OpenVoice v1 and v2. Users can either upload audio or record their voice in-browser, input custom text, select expressive voice styles or accents, and generate natural-sounding cloned audio.

---

Features

• Upload audio files or use your microphone to record directly.
• OpenVoice v1: voice cloning with style options:
  default, friendly, cheerful, excited, sad, angry, terrified, shouting, whispering
• OpenVoice v2: accent-aware voice cloning:
  American, British, Indian, Australian, Default
• Real-time audio generation and download/playback support
• Fully client-server integrated with Flask and JS (recorder.js)

---

Project Structure

.
├── app.py                   # Flask entry point
├── routes.py                # Routing logic
├── requirements.txt         # Dependencies
├── style.css                # Optional styling overrides
├── templates/
│   └── index.html           # Web UI
├── static/
│   ├── recorder.js          # Microphone recorder logic
│   ├── uploads/             # Uploaded audio storage
│   └── outputs/             # Generated audio storage
├── services/                # Core voice generation services
│   ├── audio_processing.py
│   ├── openvoice_v1.py
│   └── openvoice_v2.py
├── openvoice/               # Model-related scripts
└── checkpoints/             # Downloaded model weights
    ├── v1/
    │   ├── base_speakers/
    │   └── converter/
    └── v2/
        ├── base_speakers/
        └── converter/

---

Installation

1. Clone and Setup Environment

git clone https://github.com/anshulraj10/ai-voice-replication.git
cd ai-voice-replication
python3 -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

2. Install Requirements

pip install -r requirements.txt

3. Download Checkpoints

• OpenVoice v1: checkpoints_1226.zip
• OpenVoice v2: checkpoints_v2_0417.zip

Extract into checkpoints/ as follows:

checkpoints/
├── v1/
│   ├── base_speakers/
│   └── converter/
└── v2/
    ├── base_speakers/
    └── converter/

4. Run the App

python app.py

Then open your browser at: http://localhost:5000

---

Usage Instructions

1. Enter the text to clone.
2. Upload or record your voice.
3. Choose model version:
   • V1: pick a style.
   • V2: pick an accent.
4. Adjust speech speed if needed.
5. Click "Generate Voice" to create output.
6. Listen or download the generated audio.

---

Model Details

• OpenVoice v1 – GitHub
  • Supports expressive style-based cloning
• OpenVoice v2 – GitHub
  • Adds accent conditioning and improved quality

Both models run locally, ensuring privacy and low latency.

---

Key Technologies

• Flask — backend server
• JavaScript (MediaRecorder API) — for audio recording
• FFmpeg + Librosa + Torchaudio — audio processing
• Torch — for model inference
• dotenv — secret management

---

Contribution Guidelines

We welcome contributions! To contribute:

1. Fork this repo
2. Create a new branch (feature/your-feature)
3. Commit your changes with clear messages
4. Open a pull request with a description

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Credits & Acknowledgements

Created by Anshul Raj
Voice cloning powered by MyShell OpenVoice

---