GH Repository · modelscope
FunASR
Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
- stars
- 20,382
- 30-day movement
- +6923/day
- Related entries
- 61
- Connections
- 1
Pythonpythonfunasropenai-compatible-apimcp-serveremotion-recognitionpytorchspeech-to-textvllmtranscriptionspeech-recognitionparaformerwhisper-alternativeasrmultilingual-asrpunctuationspeaker-diarizationreal-time-asrstreaming-asrvoice-activity-detectionaudiochinese
FunASR is an open-source speech recognition toolkit covering training and inference pipelines for ASR, streaming recognition, voice activity detection, punctuation, speaker diarization, and emotion recognition. It also serves models through OpenAI-compatible and MCP interfaces.
Use it when you need a complete Python toolkit to build, train, and serve speech-to-text pipelines rather than a single pretrained model.
Use it to
- Run streaming real-time ASR pipelines
- Add punctuation and speaker diarization to transcripts
- Train custom speech recognition models
- Serve ASR via OpenAI-compatible or MCP endpoints
- Transcribe multilingual and Chinese audio
For Developers building speech-to-text applications and services
- Role
- mcp-server
- Language
- Python
- Licence
- MIT
- Forks
- 2,035
- Open issues
- 28
- Last push
- 2026-09-16
- Latest release
- v0.1.4 · 2022-12-10
topicsspeech-recognitionasrtranscriptionspeaker-diarizationstreaming-asrmcp-server