ComfyUI Chatterbox TTS & Voice Conversion Node

Custom nodes for ComfyUI that integrate the Resemble AI Chatterbox library for Text-to-Speech (TTS) and Voice Conversion (VC).

📢 Features

• Chatterbox TTS Node:
  • Synthesize speech from text.
  • Optional voice cloning using an audio prompt.
  • Adjustable parameters: exaggeration, temperature, CFG weight, seed.
• Chatterbox Voice Conversion Node:
  • Convert the voice in a source audio file to sound like a target voice.
  • Uses a target audio file for voice characteristics or defaults to a built-in voice if no target is provided.
• Automatic Model Downloading: Necessary model files are automatically downloaded from Hugging Face (ResembleAI/chatterbox) on first use if not found locally.

🎭 Chatterbox TTS Demo Samples

Check the official demo

2. Install Dependencies: Navigate to the custom node's directory and install the required packages:

   cd ComfyUI/custom_nodes/ComfyUI-Chatterbox
   pip install -r requirements.txt

3. Model Pack Directory (Automatic Setup): The node will automatically attempt to download the default model pack (resembleai_default_voice) into ComfyUI/models/chatterbox_tts/ when you first use a node that requires it. You can also manually create subdirectories in ComfyUI/models/chatterbox_tts/ and place other Chatterbox model packs there. Each pack should contain:

   • ve.pt
   • t3_cfg.pt
   • s3gen.pt
   • tokenizer.json
   • conds.pt (for default voice capabilities)

4. Restart ComfyUI.

Usage

After installation and restarting ComfyUI:

• The "Chatterbox TTS 📢" node will be available under the audio/generation category.
• The "Chatterbox Voice Conversion 🗣️" node will be available under the audio/conversion category.

Load example workflows from the workflow-examples/ directory in this repository to get started.

Notes

• The Chatterbox library is included within this custom node's src/ directory.
• Tested with Pytorch 2.7 + CUDA 12.6

Acknowledgements

• This node relies on the Chatterbox library by Resemble AI.