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
维石530125dca86b
diff --git a/runtime/python/libtorch/funasr_torch/sensevoice_bin.py b/runtime/python/libtorch/funasr_torch/sensevoice_bin.pyindex f103fea52..8de1c54c8 100644--- a/runtime/python/libtorch/funasr_torch/sensevoice_bin.py+++ b/runtime/python/libtorch/funasr_torch/sensevoice_bin.py@@ -163,12 +163,12 @@ class SenseVoiceSmall:_language_list = _language_list * Bif len(_textnorm_list) == 1 and B != 1:_textnorm_list = _textnorm_list * B-+ctc_logits, encoder_out_lens = self.ort_infer(torch.Tensor(feats).to(self.device),torch.Tensor(feats_len).to(self.device),- torch.tensor([_language_list]).to(self.device),- torch.tensor([_textnorm_list]).to(self.device),+ torch.tensor(_language_list).to(self.device),+ torch.tensor(_textnorm_list).to(self.device),)# support batch_size=1 only currentlyx = ctc_logits[0, : encoder_out_lens[0].item(), :]