# modelscope/FunASR

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

Repository: https://github.com/modelscope/FunASR
Canonical: https://ross.abutalabs.com/products/funasr
Language: Python
License: MIT
License Family: permissive
Topics: pytorch, speech-recognition, paraformer, punctuation, speaker-diarization, voice-activity-detection, asr, multilingual-asr, speech-to-text, transcription, whisper-alternative, audio, chinese, emotion-recognition, mcp-server, openai-compatible-api, streaming-asr, vllm, funasr, real-time-asr
Last push: 2026-08-26T16:55:58+00:00

## Health v2 (maintenance only)
Score: 99/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 99, longevity 98
- inputs: {"age_days": 1379, "days_push": 7, "days_rel": 7, "gap_med": 0.0, "n_releases_24m": 43}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 20036, forks 2004 (observed 2026-08-28T04:11:29.576037+00:00)

## What it is
FunASR is an industrial end-to-end speech recognition toolkit built on PyTorch, offering ASR, VAD, punctuation restoration, speaker diarization, and emotion recognition pipelines for offline, streaming, and edge deployment. It includes OpenAI-compatible serving, WebSocket streaming, vLLM acceleration, and an MCP server for integration with AI agents.

## Use cases
- transcribe audio files to text
- real-time streaming speech recognition
- add punctuation to raw transcripts
- identify speakers in meeting recordings
- run speech-to-text on Chinese and multilingual audio
- serve an OpenAI-compatible ASR API
- deploy speech recognition on edge devices

## When to choose
- you need production-grade ASR with streaming and VAD pipelines
- you want a Whisper alternative with strong Chinese/multilingual support
- you need speaker diarization and punctuation in one toolkit
- you want OpenAI-compatible or MCP-based speech serving

## When to avoid
- you only need text-to-speech synthesis
- you need a lightweight non-PyTorch dependency
- your project requires a non-MIT copyleft-compatible license

## Facets
- artifact type: library
- maturity: active
- function: speech-recognition, audio-processing, machine-learning, llm-inference, mcp, http-server
- domain: speech-processing, machine-learning
- platform: python, cross-platform, cli
- tags: asr, speech-to-text, transcription, paraformer, whisper-alternative, voice-activity-detection, speaker-diarization, punctuation, streaming-asr, openai-compatible-api, vllm, gguf, websocket, chinese, multilingual, pytorch, audio, natural-language-processing, gpu, docker

## Member repositories
- modelscope/FunASR (main) score 99

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:29.576037+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:59:22.318124+00:00, confidence not recorded.
  - readme: https://github.com/modelscope/FunASR (fetched 2026-08-28T04:11:29.576037+00:00, sha 77e6f0ae2270)
  - registry_pypi: https://pypi.org/pypi/funasr/json (fetched 2026-08-29T07:57:49.388860+00:00, sha 18ae7779e99c)
- Data as of 2026-08-30T08:39:29.467469+00:00.