Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
Commits on main
chore: fix .gitignore — use recursive patterns for __pycache__ and .DS_Store
0fc3cfc 游雁 committed 5 months ago
feat: optionally return per-speaker embedding centroids (#2967)
8541d09 phoenixray2000 committed 5 months ago
Merge pull request #2954 from modelscope/readme-add-fun-asr-nano
516c4f7 zhifu gao committed 5 months ago
release v1.3.9: add wheel packaging, fix SenseVoice+spk_model crash, torchaudio fallback
11b04b8 游雁 committed 5 months ago
fix: SenseVoice+spk_model crash and torchaudio>=2.11 compatibility
80f3f2a 游雁 committed 5 months ago
fix: remove repetition_penalty from vLLM services (fixes #2950 CUDA assert)
b8d736e 游雁 committed 5 months ago
examples: add Dify integration guide (connect via OpenAI-compatible API)
679f40c 游雁 committed 5 months ago
docs: add Chinese blog draft — FunASR vs Whisper comparison for Zhihu/CSDN
21e6ede 游雁 committed 5 months ago
feat: implement Whisper forward() for fine-tuning support (fixes #2678)
a8d23ee 游雁 committed 5 months ago
docs: add torch+torchaudio to vllm_guide_zh install section
5d8349a zhifu gao committed 5 months ago
docs: add torch+torchaudio to install instructions (not in pip deps, user must install separately)
0e9d53e zhifu gao committed 5 months ago
fix: update install_requires — add torchaudio, transformers, tiktoken, huggingface_hub, safetensors
6134188 游雁 committed 5 months ago
fix: update server version to 1.3.6, add vllm to dependency hints, add repetition_penalty to streaming
01057bf 游雁 committed 5 months ago
fix: add repetition_penalty=1.3 to streaming service (fixes #2930 streaming repetition)
91905fd 游雁 committed 5 months ago
fix: funasr-server uses vLLM engine for Fun-ASR-Nano (fixes #2930 text repetition)
787c282 游雁 committed 5 months ago
fix: add repetition_penalty=1.3 to all vLLM services (fixes #2930 text repetition)
793c45b 游雁 committed 5 months ago
fix: vLLM >=0.21 compat — EmbedsPrompt moved from vllm.inputs.data to vllm.inputs
3953072 游雁 committed 5 months ago