Whissle

Provides access to Whissle API endpoints for speech-to-text, diarization, translation, and text summarization.

python

0

GitHub Stars

python

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "mcp-mirror-whissleai_whissle-mcp": {
      "command": "python",
      "args": [
        "/path/to/whissle_mcp/server.py"
      ],
      "env": {
        "WHISSLE_AUTH_TOKEN": "insert_auth_token_here",
        "WHISSLE_MCP_BASE_PATH": "/path/to/your/base/directory"
      }
    }
  }
}

You run a Python-based MCP server that exposes Whissle API endpoints for speech-to-text, diarization, translation, and text summarization. This server lets you integrate Whissle capabilities into your workflows and client applications with a simple local or remote execution path.

How to use

To use the Whissle MCP Server, start the local stdio-based server configuration and point your MCP client at it. The server handles transcription, diarization, translation, and summarization via the Whissle API. You will interact with it through the MCP client’s usual request/response flow, using the server as the backend that processes audio or text and returns structured results.

How to install

Prerequisites: you need Python 3.8 or higher and pip installed on your system.

Create a project directory and navigate into it.
Create and activate a virtual environment.
Install the MCP package in editable mode.
Create a local environment file with authentication and base path settings.
Configure the integration entry point for MCP. You will point the MCP runner at your Python interpreter inside the virtual environment and the server script path.

Configuration and usage notes

Environment variables you will use with the local MCP setup include the Whissle API authentication token and a base path for file operations.

Supported audio formats include WAV, MP3, OGG, FLAC, and M4A. The maximum file size is 25 MB. Follow good practices for handling API tokens by keeping them out of version control.

The server provides tools for speech-to-text, speech diarization, text translation, text summarization, and listing available ASR models. Each tool can be invoked via your MCP client to obtain transcripts, diarized outputs, translations, or summaries.

Security and troubleshooting

Keep your authentication token secret and rotate it as needed. Validate audio file validity, size, and format before sending requests. If you encounter errors, check token validity, file constraints, and network connectivity. The server includes error handling and retry mechanisms for transient failures.

Examples and best practices

Split long audio into smaller chunks when possible to balance latency and cost. Use word timestamps when you need precise alignment with transcripts. When translating or summarizing, consider preprocessing the text to remove noise for better results.

Available tools

speech_to_text

Convert speech to text from an audio file using the Whissle API. Supports options for timestamps and boosted vocabulary to improve recognition for specific terms.

diarize_speech

Perform speech-to-text with speaker diarization to identify which speaker is speaking when transcribing audio.

translate_text

Translate text from a source language to a target language using the Whissle translation endpoint.

summarize_text

Summarize long text content using a language model, with optional instruction to guide the summary style.

list_asr_models

List all available automatic speech recognition models and their capabilities.