Cloud ASR MCP

A Model Context Protocol (MCP) server providing cloud-based speech-to-text transcription using multiple backends: Google Gemini, OpenAI, and AssemblyAI.

Features

• 5 transcription tools across 3 cloud backends
• Google Gemini: Raw and cleaned transcription modes with LLM-powered editing
• OpenAI: Standard (gpt-4o-transcribe) and economy (gpt-4o-mini-transcribe) options
• AssemblyAI: Job-based workflow ideal for long-form audio
• Automatic file handling: Large file downsampling via ffmpeg
• Optional file output: Save transcripts as markdown files

Tools

| Tool | Backend | Description | |------|---------|-------------| | gemini_transcribe | Gemini | Lightly edited transcript with filler words removed, punctuation added | | gemini_transcribe_raw | Gemini | Verbatim transcript including filler words and corrections | | openai_transcribe | OpenAI | High-quality transcription with gpt-4o-transcribe | | openai_transcribe_economy | OpenAI | Cost-effective option using gpt-4o-mini-transcribe | | assemblyai_transcribe | AssemblyAI | Best for long-form content with job-based processing |

Installation

From npm

npm install -g cloud-asr-mcp

From source

git clone https://github.com/danielrosehill/Cloud-ASR-MCP.git
cd Cloud-ASR-MCP/mcp
npm install
npm run build

Configuration

Set the following environment variables for the backends you want to use:

GEMINI_API_KEY="your-gemini-api-key"
OPENAI_API_KEY="your-openai-api-key"
ASSEMBLYAI_API_KEY="your-assemblyai-api-key"  # Also accepts ASSEMBLY_API_KEY

Claude Code MCP Configuration

Add to your Claude Code MCP settings (~/.claude/settings.json):

Using npm package (recommended)

{
  "mcpServers": {
    "cloud-asr": {
      "command": "npx",
      "args": ["-y", "cloud-asr-mcp"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key",
        "OPENAI_API_KEY": "your-openai-api-key",
        "ASSEMBLYAI_API_KEY": "your-assemblyai-api-key"
      }
    }
  }
}

Using local installation

{
  "mcpServers": {
    "cloud-asr": {
      "command": "node",
      "args": ["/path/to/Cloud-ASR-MCP/mcp/dist/index.js"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key",
        "OPENAI_API_KEY": "your-openai-api-key",
        "ASSEMBLYAI_API_KEY": "your-assemblyai-api-key"
      }
    }
  }
}

Usage Examples

Basic transcription

Use gemini_transcribe to transcribe /path/to/audio.mp3

Save transcript to file

Transcribe /path/to/audio.mp3 using assemblyai_transcribe and save to /home/user/transcripts/

OpenAI with prompt for better accuracy

Use openai_transcribe on /path/to/meeting.wav with prompt "Technical discussion about Kubernetes, Docker, and CI/CD pipelines"

Supported Audio Formats

• MP3, WAV, OGG, FLAC, AAC, AIFF, M4A, WEBM, MPEG

File Size Limits

| Backend | Max Size | |---------|----------| | OpenAI | 25 MB | | Gemini | 100 MB (auto-downsampled if >15 MB) | | AssemblyAI | 5 GB |

Requirements

• Node.js >= 18.0.0
• ffmpeg (for audio downsampling of large files)

License

MIT

Links

• GitHub Repository
• npm Package
• Report Issues