Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@mistral-ai-mcp ocr this pdf: https://example.com/doc.pdf"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Mistral AI MCP server + CLI. OCR documents, TTS text-to-speech, STT speech-to-text via Voxtral.
OCR - Extract text from PDFs, DOCX, PPTX, XLSX, images
TTS - Generate speech with preset voices or voice cloning
STT - Transcribe audio (batch or realtime)
Related MCP server: simple-asr-mcp
npm install -g mistral-ai-mcpRequires Node.js 18+.
mistral-ai ocr <file-or-url> # Extract text from documents/images
mistral-ai tts <text> # Generate speech from text
mistral-ai stt <audio> # Transcribe audio to text
mistral-ai config ... # Manage configuration# Local PDF
mistral-ai ocr ./document.pdf > output.md
# From URL
mistral-ai ocr https://arxiv.org/pdf/2301.00001.pdf > paper.md
# Tables
mistral-ai ocr ./document.pdf --table-format html# Preset voice
mistral-ai tts "Hello, world!" --voice-id alice
# Voice cloning via reference audio
mistral-ai tts "Hello from me!" --ref-audio ./my-voice.wav
# Output format
mistral-ai tts "Hello!" --voice-id bob --format wav > output.wav# Basic transcription
mistral-ai stt ./audio.mp3
# Realtime mode (low latency)
mistral-ai stt ./audio.mp3 --realtime
# Speaker diarization
mistral-ai stt ./meeting.mp3 --diarize
# Specific language
mistral-ai stt ./audio.mp3 --language enTool | Input Formats | Output Formats |
OCR | PDF, DOCX, PPTX, XLSX, PNG, JPEG, AVIF | Markdown + YAML |
TTS | Text | MP3, WAV, PCM, FLAC, Opus |
STT | MP3, WAV, FLAC, OGG, WebM | Text + JSON |
mistral-ai config api_key <your-key>Or via environment:
export MISTRAL_API_KEY=your-keyLinux/macOS: ~/.mistral-ai/config.json
Windows: %USERPROFILE%\.mistral-ai\config.json
Override with MISTRAL_AI_CONFIG_DIR env var.
mistral-ai config showmistral-ai config api_key <key> # Set API key
mistral-ai config base_url <url> # API endpoint (default: https://api.mistral.ai/v1)
mistral-ai config model <model> # Default OCR modelStart server with stdio transport:
mistral-ai-mcpTool | Description |
| Extract text from PDF sources |
| Generate speech from text |
| Transcribe audio to text |
import { StdioClientTransport } from '@modelcontextprotocol/sdk/client/stdio.js';
import { Client } from '@modelcontextprotocol/sdk/client/index.js';
const transport = new StdioClientTransport({
command: 'node',
args: ['node_modules/.bin/mistral-ai-mcp'],
env: { MISTRAL_API_KEY: 'your-key' },
});
const client = new Client({ name: 'test', version: '1.0.0' }, { capabilities: {} });
await client.connect(transport);
// OCR
const ocrResult = await client.callTool({
name: 'ocr_pdf_url',
arguments: { pdf_url: 'https://example.com/doc.pdf' },
});
// TTS
const ttsResult = await client.callTool({
name: 'tts_speech',
arguments: { text: 'Hello!', voice_id: 'alice' },
});
// STT
const sttResult = await client.callTool({
name: 'stt_transcribe',
arguments: { audio_source: './audio.mp3' },
});npm installnpm run dev # Run with tsx (watch mode)
npm run build # Compile TypeScript
npm run clean # Remove dist/
npm run lint # ESLint check
npm run format # Prettier check
npm run format:write # Prettier fix
npm run typecheck # Type check only
npm run test # Run testsHusky hooks enforce code quality:
pre-commit: Runs npm run lint && npm run format
pre-push: Runs npm run typecheck && npm run test
GitHub Actions workflows:
CI (.github/workflows/ci.yml): Lint, build, typecheck on push/PR
Tests (.github/workflows/test.yml): Run tests on push/PR
ISC
This server cannot be deployed
MCP server for Speech-to-Text
MCP server for Text-to-Speech
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
OCR, transcription, file extraction, and image generation for AI agents via MCP.