🗣️ Chatterbox TTS Setup

🔗 Here is the link of Google Colab link : https://colab.research.google.com/drive/1ghLnU6tz1QT1685QhjptIF-lPG_87aie?usp=sharing

A streamlined, battle-tested setup for Chatterbox TTS — the expressive, voice-cloning text-to-speech model by Resemble AI. This repository provides two foolproof ways to get Chatterbox running:

• ☁️ Google Colab Notebook – Zero local setup, runs entirely in the cloud with GPU acceleration. 🔗 Here is the link of Google Colab link : https://colab.research.google.com/drive/1ghLnU6tz1QT1685QhjptIF-lPG_87aie?usp=sharing
• 💻 Local Python Script – For users who want to run Chatterbox on their own machine.

Both methods include:

• Automatic dependency resolution (solves the tricky torch.fx / torchao conflicts).
• Voice cloning from a short reference audio file.
• Adjustable expressiveness (exaggeration, cfg_weight).
• Automatic text chunking for long passages.
• Audio playback and waveform visualization.

---

🚀 Quick Start

☁️ Option 1: Google Colab (Recommended)

Click the badge above or open this link.
Run the cells in order – the notebook will install everything, download the model, and generate speech.
Upload a voice sample to your Google Drive for cloning (optional).

💻 Option 2: Local Setup

git clone https://github.com/Yonatankinfe/chatterbox-tts.git
cd chatterbox-tts
python local_chatterbox.py

The script will:

• Create a virtual environment (optional).
• Install exact dependencies.
• Download the model.
• Generate a sample TTS output.

###⚠️ System Requirements (Local)

• Python 3.10 or 3.11 (3.12 may work but is untested).
• CUDA‑compatible GPU recommended (CPU fallback works, slowly).
• ~6 GB free disk space for model weights.

📦 Dependencies

torch==2.5.0
torchaudio==2.5.0
transformers==4.46.3
diffusers==0.29.0
protobuf==3.20.3
chatterbox-tts (latest, --no-deps)
resemble-perth
s3tokenizer
conformer
librosa==0.11.0

📁 Repository Structure

chatterbox-tts/
├── README.md                 # You are here
├── chatterbox_colab.ipynb    # Google Colab notebook (end‑to‑end)
└── local_chatterbox.py       # Single‑script local setup + demo

🎛️ Configuration Options

Both the notebook and local script expose key parameters:

🎭 Presets

The code includes several speaking‑style presets:

"neutral", "calm", "expressive", "dramatic", "storytelling", "audiobook", "fast_speaker"

Example

config.get_preset("storytelling")  # Returns dict with exaggeration=0.8, cfg_weight=0.4

Example

🎤🎤 Sample voice

🎤 download (8).wav

🎤🎤Sample voice for content Creation

🎤 download (7).wav

Sample image

### ⚠️ Known Limitations

• Voice Samples: Must be WAV format, 10-30 seconds, clean audio
• Long Texts: Very long passages (>2000 words) may require multiple runs
• GPU Memory: 6GB+ VRAM recommended for optimal performance
• Python Version: Tested on 3.10/3.11; 3.12 untested

🙏 Acknowledgments

• Resemble AI for creating and open-sourcing Chatterbox
• The PyTorch team for the incredible deep learning framework
• All early testers who helped identify edge cases