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
bug fix
shixian.shi0c64e61c97a8
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.pyindex e4a154dbb..f59bb6b7a 100644--- a/funasr/auto/auto_model.py+++ b/funasr/auto/auto_model.py@@ -400,18 +400,18 @@ class AutoModel:for res, vadsegment in zip(restored_data, vadsegments):sentence_list.append({"start": vadsegment[0],\"end": vadsegment[1],- "sentence": res['text'],+ "sentence": res['raw_text'],"timestamp": res['timestamp']})elif self.spk_mode == 'punc_segment':sentence_list = timestamp_sentence(punc_res[0]['punc_array'], \result['timestamp'], \- result['text'])+ result['raw_text'])distribute_spk(sentence_list, sv_output)result['sentence_info'] = sentence_listelif kwargs.get("sentence_timestamp", False):sentence_list = timestamp_sentence(punc_res[0]['punc_array'], \result['timestamp'], \- result['text'])+ result['raw_text'])result['sentence_info'] = sentence_listif "spk_embedding" in result: del result['spk_embedding']