Voice Recognition

You can run this Voice Recognition MCP Server locally to perform voice recognition and text extraction via lightweight stdio or through an MCP client. It supports decoding from audio files or base64 data and returns structured results you can use in further processing.

How to use

You use a local MCP client to connect to the server in either stdio or MCP mode. In stdio mode, you launch the service and feed requests through standard input, then read structured responses from standard output. In MCP mode, you run the MCP entry point and communicate using the MCP protocol as your transport. The service will process voice data or text and return results that include the recognized text and metadata such as language and speaker information.

How to install

Prerequisites: you need Python installed on your system. You will also install project dependencies listed in a requirements file.

Step 1: Prepare your environment by creating a working directory and navigating into it.

Step 2: Install dependencies.

pip install -r requirements.txt

Run in stdio mode

To run the stdio mode service, start the stdio entry point and then interact through stdin/stdout.

Alternatively, use the prebuilt executable for stdio mode if you prefer a binary.

Run in MCP mode

To run the MCP mode service, start the MCP entry point. You can run it with Python for development or use the prebuilt binary for production.

Configuration and environment

Set up environment variables to configure the API access used by the service. The following variables are used as placeholders until you provide real values:

API_URL: your_api_url API_KEY: your_api_key

Notes on usage and results

The service returns structured results that include the recognized text and a metadata block with language, emotion, audio type, speaker, and the raw text. You can use these fields to drive downstream processing or display results to users.

Troubleshooting and tips

If the server cannot start, check that Python is installed and that dependencies are installed. Ensure the environment variables API_URL and API_KEY are set correctly. If you switch between stdio and MCP modes, make sure you start the correct entry point corresponding to your chosen transport.

Security considerations

Keep your API key secret. Do not commit credentials into code or logs. If deploying in production, consider restricting access to the local MCP port and rotating API keys periodically.

Examples and capabilities

This service can recognize voice from a file, decode base64-encoded audio, and extract text. It responds with a consistent, structured payload that can be parsed by your application.

Migration notes

If you upgrade to a newer build, verify that the MCP/stdio entry points continue to start as before and that environment variables remain correctly configured.

Support and吧

If you encounter issues that are not covered here, prepare a minimal reproduction and check the logs produced by the service to identify configuration or data issues.

Server configurations (MCP endpoints)

[
  {
    "type": "http",
    "name": "voiceidentify_http",
    "url": "https://localhost:8000/mcp",
    "args": []
  },
  {
    "type": "stdio",
    "name": "voice_stdio_local",
    "command": "python",
    "args": ["stdio_server.py"]
  },
  {
    "type": "stdio",
    "name": "voice_mcp_local",
    "command": "./dist/voice_mcp",
    "args": []
  }
]

Tooling and capabilities

The service exposes a small set of core operations for end-to-end processing of voice data and text extraction. You can identify voice from a file, identify voice from base64-encoded data, and extract text from input data.

Appendix: response formats

Responses follow a structured format with a top-level jsonrpc field and either an output or result payload depending on the response type. The voice response includes a text field and a metadata block describing language and other attributes.

Available tools

identify_voice

Identify voice from a file and return the recognized text along with metadata such as language and speaker.

identify_voice_base64

Identify voice from base64 encoded data and return the recognized text with metadata.

extract_text

Extract text from provided input and return it in a structured format.