mcp-stt-whisper mcp

v1.0.1 · MCP Tool · audio · registry.pascalai.org

Speech-to-text via local Whisper.cpp server (OpenAI-compatible API). Fast, offline, no API key required. Runs at http://localhost:8080 by default. Compatible with whisper.cpp --server and faster-whisper-server.

sttwhisperlocalofflinetranscriptionspeech

Input Parameters

Parameter	Type	Description
filePathrequired	string	Path to the audio file (mp3, wav, ogg, m4a, flac, webm).
languageoptional	string	Language code (e.g. es, en, fr). Empty = auto-detect. Default: .
operationoptional	string	Operation: transcribe (default) \| translate (to English). Default: `transcribe`.
promptoptional	string	Context hint to guide transcription style.
modeloptional	string	Model name (e.g. base, small, medium, large-v3). Server decides if not specified. Default: `base`.
apiKeyoptional	string	Unused — Whisper.cpp runs locally without authentication.
baseUrloptional	string	Whisper server URL (default: http://localhost:8080). Default: `http://localhost:8080`.

Output Fields

Field	Type	Description
text	string	Transcribed text.
language	string
duration	number
operation	string
error	string

Examples

Transcribe local audio with Whisper

// Input
{
  "filePath": "C:\\recordings\\lecture.mp3",
  "model": "medium",
  "language": "en"
}

// Output
{
  "text": "Today we will discuss the fundamentals of neural networks.",
  "language": "en",
  "duration": 0
}

Translate Spanish audio to English

// Input
{
  "filePath": "C:\\audio\\clase.mp3",
  "operation": "translate"
}

// Output
{
  "text": "The student asked a very interesting question.",
  "language": "",
  "duration": 0
}

Install & Discovery

Install

ppm install mcp-stt-whisper

Get JSON Schema

GET /v1/packages/mcp-stt-whisper/1.0.1/schema

Discover by keyword

GET /v1/mcp/discover?q=stt

Discovery hint: Install with ppm install mcp-stt-whisper or invoke remotely via POST /v1/invoke/mcp-stt-whisper on the MCP Service.