MCP server

The Model Context Protocol (MCP) lets an assistant call tools directly. The Speech Revolutions MCP server exposes this API as five tools, so you can ask an assistant to “transcribe this recording and tell me who said what” without writing code or copying job IDs.

The server ships inside the JavaScript SDK. There is nothing separate to install.

Setup

Add this to your MCP client's configuration. In Claude Desktop that is claude_desktop_config.json; other clients use the same shape.

json
{
  "mcpServers": {
    "speechrevolutions": {
      "command": "npx",
      "args": ["-y", "speechrevolutions", "mcp"],
      "env": {
        "SPEECHREVOLUTIONS_API_KEY": "stt_..."
      }
    }
  }
}

Restart the client. If you don't have an API key, create one in the console. New accounts start with $10 of credit, about 55 hours of audio.

The server runs locally as a subprocess of your MCP client and reads the API key from its environment. The key is sent only to the Speech Revolutions API, the same way the SDK sends it.

Tools

ToolWhat it does
transcribe_audioTranscribe a local file or public URL and return the text. Waits for the result, so use it for recordings up to about 30 minutes.
submit_transcription_jobSubmit a recording and return a job ID immediately. Use it for longer recordings and for batches.
check_jobReport whether a job is processing, complete, or failed.
get_transcriptReturn a completed transcript as text, JSON, SRT, or VTT.
list_jobsList recent jobs on the account, newest first.

Output

Transcripts are returned as readable text with speaker labels and timestamps, not as raw JSON:

text
Job: f1c3ae59-00bc-4e39-aa9e-5afed8c3974a · Detected language: en · Speakers: 3

[0:00] SPEAKER_1: You are already ahead of 90% of the people your age if you
can just communicate competently.
[0:12] SPEAKER_2: That tracks with what we saw in the hiring data.

This keeps long recordings within the assistant's context. To get per-word timings and confidences, call get_transcript with format: "json".

Long transcripts are truncated with a notice that names the job ID and how much was cut. Raise max_characters to get more.

Cost

The same as any other API call: $0.003 per minute of audio, with diarization and timestamps included. Tool calls that only check status or list jobs are free.

Troubleshooting

If the server doesn't appear in your client, run it manually. It prints errors to stderr, which some clients hide:

bash
SPEECHREVOLUTIONS_API_KEY=stt_... npx -y speechrevolutions mcp

It prints a readiness line and then waits for input. If it reports that the key is unset, the env block in your configuration isn't reaching the process.