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
update device
shixian.shieed0c2381691
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.pyindex 107c78ecb..0641f06a8 100644--- a/funasr/auto/auto_model.py+++ b/funasr/auto/auto_model.py@@ -398,7 +398,7 @@ class AutoModel:if self.spk_model is not None:all_segments = sorted(all_segments, key=lambda x: x[0])spk_embedding = result['spk_embedding']- labels = self.cb_model(spk_embedding, oracle_num=self.preset_spk_num)+ labels = self.cb_model(spk_embedding.cpu(), oracle_num=self.preset_spk_num)del result['spk_embedding']sv_output = postprocess(all_segments, None, labels, spk_embedding.cpu())if self.spk_mode == 'vad_segment':