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
batch
游雁b512846c2ca0
diff --git a/funasr/models/sense_voice/model.py b/funasr/models/sense_voice/model.pyindex 07fb4eb58..ae20902cf 100644--- a/funasr/models/sense_voice/model.py+++ b/funasr/models/sense_voice/model.py@@ -329,6 +329,8 @@ class SenseVoiceRWKV(nn.Module):stats["loss"] = torch.clone(loss.detach())stats["batch_size"] = batch_sizestats["batch_size_x_frames"] = frames * batch_size+ stats["batch_size_real_frames"] = speech_lengths.sum().item()+ stats["padding_frames"] = stats["batch_size_x_frames"] - stats["batch_size_real_frames"]# force_gatherable: to-device and to-tensor if scalar for DataParallelif self.length_normalized_loss: