Voice Recognition MCP Service

This service provides voice recognition and text extraction capabilities through both stdio and MCP modes.

Features

Voice recognition from file
Voice recognition from base64 encoded data
Text extraction
Support for both stdio and MCP modes
Structured voice recognition results

Project Structure

voice_service.py - Core service implementation
stdio_server.py - stdio mode entry point
mcp_server.py - MCP mode entry point
build.py - Build script for executables
build_exec.sh - Build execution script
test_*.sh - Test scripts for different functionalities

Installation

Clone the repository:

git clone https://github.com/AIO-2030/mcp_voice_identify.git
cd mcp_voice_identify

Install dependencies:

pip install -r requirements.txt

Set up environment variables in .env:

API_URL=your_api_url
API_KEY=your_api_key

Usage

stdio Mode

Run the service:

python stdio_server.py

Send JSON-RPC requests via stdin:

{
    "jsonrpc": "2.0",
    "method": "help",
    "params": {},
    "id": 1
}

Or use the executable:

./dist/voice_stdio

MCP Mode

Run the service:

python mcp_server.py

Or use the executable:

./dist/voice_mcp

Voice Recognition Results

The service provides structured voice recognition results. Here's an example of the response format:

Original API Response

{
    "jsonrpc": "2.0",
    "result": {
        "message": "input processed successfully",
        "results": "test test test",
        "label_result": "<|en|><|EMO_UNKNOWN|><|Speech|><|woitn|>test test test"
    },
    "id": 1
}

Restructured Response

{
    "jsonrpc": "2.0",
    "result": {
        "message": "input processed successfully",
        "results": "test test test",
        "label_result": {
            "lan": "en",
            "emo": "unknown",
            "type": "speech",
            "speaker": "woitn",
            "text": "test test test"
        }
    },
    "id": 1
}

Label Result Fields

The label_result field contains the following structured information:

Field	Description	Example Value
lan	Language code	"en"
emo	Emotion state	"unknown"
type	Audio type	"speech"
speaker	Speaker identifier	"woitn"
text	Recognized text content	"test test test"

Special Labels

The service recognizes and processes the following special labels in the original response:

<|en|> - Language code
<|EMO_UNKNOWN|> - Emotion state
<|Speech|> - Audio type
<|woitn|> - Speaker identifier

Building Executables

Make the build script executable:

chmod +x build_exec.sh

Build stdio mode executable:

./build_exec.sh

Build MCP mode executable:

./build_exec.sh mcp

The executables will be created at:

stdio mode: dist/voice_stdio
MCP mode: dist/voice_mcp

Testing

Run the test scripts:

chmod +x test_*.sh
./test_help.sh
./test_voice_file.sh
./test_voice_base64.sh

License

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