alltalk-tts-mcp

A Model Context Protocol (MCP) server for AllTalk TTS, providing full-featured text-to-speech with RVC voice conversion, narrator mode, and advanced audio generation capabilities.

• Full AllTalk API Support: All parameters from SillyTavern's AllTalk integration
• RVC Voice Conversion: Real-Time Voice Cloning with GPU acceleration
• Narrator Mode: Separate voices for character dialogue and narration
• Multiple Output Formats: WAV audio with configurable parameters
• Performance Optimizations: DeepSpeed and Low VRAM modes
• Cross-platform Audio Playback: Automatic audio playback after generation

1. AllTalk TTS Server: Running locally or remotely

• Default: http://localhost:7851
• AllTalk GitHub

1. Python 3.8+ with virtual environment support
2. Audio System: PulseAudio, ALSA, or compatible audio backend

git clone https://github.com/yourusername/alltalk-tts-mcp.git
cd alltalk-tts-mcp

# Install uv if not already installed
curl -LsSf https://astral.sh/uv/install.sh | sh
# Or using pip: pip install uv

# Create virtual environment and install dependencies
uv sync

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt

Create a .env file or set environment variables:

# Required: AllTalk server URL
export ALLTALK_URL="http://localhost:7851"

# Optional: Audio playback settings
export PULSE_SERVER="unix:/run/user/$(id -u)/pulse/native"
export XDG_RUNTIME_DIR="/run/user/$(id -u)"

{
  "mcpServers": {
    "alltalk-tts":