# voiceroid_daemon-mcp

MCP (Model Context Protocol) server for VOICEROID2 text-to-speech via [voiceroid_daemon](https://github.com/Nkyoku/voiceroid_daemon).

## Features

- Text-to-speech generation with VOICEROID2 voices
- Text-to-kana phonetic conversion
- Customizable voice parameters (volume, speed, pitch, emphasis)
- Audio playback support for macOS, Windows, and Linux
- Basic authentication support

## Prerequisites

- Node.js 18 or higher
- [voiceroid_daemon](https://github.com/Nkyoku/voiceroid_daemon) running on your system
- VOICEROID2 installed (for voiceroid_daemon)

## Installation

```bash
# Clone the repository
git clone https://github.com/mohemohe/voiceroid_daemon-mcp.git
cd voiceroid_daemon-mcp

# Install dependencies
npm install
```

## Configuration

Create a `.env` file in the project root (optional):

```env
# voiceroid_daemon server URL (default: http://127.0.0.1:8080)
VOICEROID_DAEMON_URL=http://127.0.0.1:8080

# Basic authentication (if required)
VOICEROID_DAEMON_USERNAME=your_username
VOICEROID_DAEMON_PASSWORD=your_password

# Default voice parameters (optional)
VOICEROID_DEFAULT_VOLUME=1.0        # 0-2
VOICEROID_DEFAULT_SPEED=1.3         # 0.5-4
VOICEROID_DEFAULT_PITCH=1.0         # 0.5-2
VOICEROID_DEFAULT_EMPHASIS=1.1      # 0-2
VOICEROID_DEFAULT_PAUSE_MIDDLE=150  # 80-500
VOICEROID_DEFAULT_PAUSE_LONG=370    # 100-2000
VOICEROID_DEFAULT_PAUSE_SENTENCE=800 # 0-10000
```

## Usage

### Running the MCP Server

```bash
# Run directly with tsx (no build required)
npm start

# Or for development with auto-reload
npm run dev
```

### Configuring with Claude Desktop

Add the following to your Claude Desktop configuration file:

**macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
**Windows**: `%APPDATA%\Claude\claude_desktop_config.json`

```json
{
  "mcpServers": {
    "voiceroid-daemon": {
      "command": "npx",
      "args": ["tsx", "/path/to/voiceroid_daemon-mcp/src/index.ts"],
      "env": {
        "VOICEROID_DAEMON_URL": "http://127.0.0.1:8080"
      }
    }
  }
}
```

## Available Tools

### test_connection

Test the connection to voiceroid_daemon server.

```
No parameters required
```

### convert_text

Convert Japanese text to phonetic kana reading.

Parameters:
- `text` (string, required): Text to convert to kana

### speak_text

Generate and play speech audio from text.

Parameters:
- `text` (string, required): Text to speak
- `kana` (string, optional): Phonetic reading in kana
- `volume` (number, optional): Voice volume (0-2, default: 1)
- `speed` (number, optional): Speech speed (0.5-4, default: 1)
- `pitch` (number, optional): Voice pitch (0.5-2, default: 1)
- `emphasis` (number, optional): Emphasis level (0-2, default: 1)

## Example Usage in Claude

Once configured, you can use the tools in Claude:

```
Use the test_connection tool to check if voiceroid_daemon is running.

Convert "こんにちは" to kana using the convert_text tool.

Use speak_text to say "こんにちは、今日はいい天気ですね" with speed 1.2.
```

## Troubleshooting

### Connection Failed

1. Ensure voiceroid_daemon is running
2. Check the URL in your configuration
3. Verify firewall settings allow connections
4. Test with curl: `curl http://127.0.0.1:8080/`

### Audio Playback Issues

- **macOS**: Uses `afplay` (built-in)
- **Windows**: Uses PowerShell's `Media.SoundPlayer`
- **Linux**: Requires `aplay` (usually part of alsa-utils)

### Authentication Errors

If voiceroid_daemon requires authentication, ensure you've set:
- `VOICEROID_DAEMON_USERNAME`
- `VOICEROID_DAEMON_PASSWORD`

## Development

```bash
# Type checking
npm run typecheck

# Linting
npm run lint

# Run in development mode
npm run dev
```

## License

MIT