Voice Recorder MCP Server

What is Voice Recorder MCP Server?

The Voice Recorder MCP Server records audio from a microphone and transcribes it using OpenAI's Whisper model. It works as a custom extension for the Goose AI agent or as a standalone MCP server.

How to use Voice Recorder MCP Server?

Install from source with git clone and pip install -e ., then run voice-recorder-mcp with optional --model and --sample-rate arguments. Configure via environment variables like WHISPER_MODEL, SAMPLE_RATE, and MAX_DURATION. Use the tools start_recording, stop_and_transcribe, and record_and_transcribe to manage recordings.

Key features of Voice Recorder MCP Server

• Records audio from the default microphone
• Transcribes recordings using OpenAI Whisper
• Integrates with Goose AI agent as a custom extension
• Supports multiple Whisper model sizes (tiny.en, base.en, small.en, medium.en, large)
• Configurable sample rate and maximum recording duration
• Includes prompts for common recording scenarios

Use cases of Voice Recorder MCP Server

• Quickly record and transcribe voice notes or meeting snippets
• Automate transcription tasks within a Goose AI agent workflow
• Dictate commands or calculations and have the agent act on the transcription
• Compare accuracy and speed across different Whisper models
• Test microphone audio capture and transcription in a controlled environment

FAQ from Voice Recorder MCP Server

What Whisper models can I use?

The server supports tiny.en, base.en, small.en, medium.en, and large. Models with a .en suffix are optimized for English, offering faster and more accurate transcription.

How do I integrate with Goose?

In Goose, go to Settings > Extensions > Add > Command Line Extension, set the name to voice-recorder, and provide the full path to the voice-recorder-mcp executable.

What are the system requirements?

Python 3.12 or later and an audio input device (microphone).

How can I adjust recording quality?

Set the sample rate via the --sample-rate command-line argument or the SAMPLE_RATE environment variable (default 16000). Use higher sample rates for better audio quality, though file sizes increase.

What if no audio is recorded?

Check your microphone permissions and settings. Also ensure the server has access to the audio device. If model download fails, verify your internet connection.