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
feat: add Qwen3-ASR streaming WebSocket server with notes (#3035)
8f8ce5f qiulang committed 4 months ago
feat(AutoModelVLLM): warn when a single audio input is too long to decode in one pass (#3033)
2499159 zhifu gao committed 4 months ago
feat(fun_asr_nano streaming): bound the partial re-decode window (--partial-window-sec) (#3032)
f17ed67 zhifu gao committed 4 months ago
docs(README i18n): make Why-FunASR honest and Nano-first (zh/ja/ko)
2662e5c LauraGPT committed 4 months ago
docs(README): fix tagline speed claim (170x realtime was mislabeled "170x faster than Whisper")
319643f LauraGPT committed 4 months ago
docs(README): restructure into a clean get-started flow (rebased on main) (#3022)
40470b1 zhifu gao committed 4 months ago
fix(paraformer): place cif_v1_export empty-fire guard after fire_idxs is defined
57f9777 LauraGPT committed 4 months ago
fix(paraformer): guard CIF against IndexError when no boundaries fire (#3028)
1da63a4 zhifu gao committed 4 months ago
fix(sense_voice): remove docstring between @triton.jit and def (SyntaxError) (#3027)
e79afd5 zhifu gao committed 5 months ago
docs(i18n README): align benchmark with canonical (add Chinese CER, flagship Nano 340x) (#3025)
352daed zhifu gao committed 5 months ago
docs(README): add Chinese CER to benchmark (canonical micro-avg, consistent with funasr.com) (#3024)
f4cab0d zhifu gao committed 5 months ago