Voice Recorder (Whisper)

An MCP server for recording audio and transcribing it using OpenAI's Whisper model. Designed to work as a Goose custom extension or standalone MCP server.

Features

• Record audio from the default microphone
• Transcribe recordings using Whisper
• Integrates with Goose AI agent as a custom extension
• Includes prompts for common recording scenarios

Installation

# Install from source
git clone https://github.com/DefiBax/voice-recorder-mcp.git
cd voice-recorder-mcp
pip install -e .

Usage

As a Standalone MCP Server

# Run with default settings (base.en model)
voice-recorder-mcp

# Use a specific Whisper model
voice-recorder-mcp --model medium.en

# Adjust sample rate
voice-recorder-mcp --sample-rate 44100

Testing with MCP Inspector

The MCP Inspector provides an interactive interface to test your server:

# Install the MCP Inspector
npm install -g @modelcontextprotocol/inspector

# Run your server with the inspector
npx @modelcontextprotocol/inspector voice-recorder-mcp

With Goose AI Agent

1. Open Goose and go to Settings > Extensions > Add > Command Line Extension

2. Set the name to voice-recorder

3. In the Command field, enter the full path to the voice-recorder-mcp executable:

   /full/path/to/voice-recorder-mcp
   

   Or for a specific model:

   /full/path/to/voice-recorder-mcp --model medium.en
   

   To find the path, run:

   which voice-recorder-mcp
   

4. No environment variables are needed for basic functionality

5. Start a conversation with Goose and introduce the recorder with: "I want you to take action from transcriptions returned by voice-recorder. For example, if I dictate a calculation like 1+1, please return the result."

Available Tools

• start_recording: Start recording audio from the default microphone
• stop_and_transcribe: Stop recording and transcribe the audio to text
• record_and_transcribe: Record audio for a specified duration and transcribe it

Whisper Models

This extension supports various Whisper model sizes:

Model	Speed	Accuracy	Memory Usage	Use Case
tiny.en	Fastest	Lowest	Minimal	Testing, quick transcriptions
base.en	Fast	Good	Low	Everyday use (default)
small.en	Medium	Better	Moderate	Good balance
medium.en	Slow	High	High	Important recordings
large	Slowest	Highest	Very High	Critical transcriptions

The .en suffix indicates models specialized for English, which are faster and more accurate for English content.

Requirements

• Python 3.12+
• An audio input device (microphone)

Configuration

You can configure the server using environment variables:

# Set Whisper model
export WHISPER_MODEL=small.en

# Set audio sample rate
export SAMPLE_RATE=44100

# Set maximum recording duration (seconds)
export MAX_DURATION=120

# Then run the server
voice-recorder-mcp

Troubleshooting

Common Issues

• No audio being recorded: Check your microphone permissions and settings
• Model download errors: Ensure you have a stable internet connection for the initial model download
• Integration with Goose: Make sure the command path is correct
• Audio quality issues: Try adjusting the sample rate (default: 16000)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

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