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
add
游雁27d2d3ccb7f6
diff --git a/examples/industrial_data_pretraining/sense_voice/demo.py b/examples/industrial_data_pretraining/sense_voice/demo.pyindex 46ec93059..b2fca4749 100644--- a/examples/industrial_data_pretraining/sense_voice/demo.py+++ b/examples/industrial_data_pretraining/sense_voice/demo.py@@ -17,6 +17,7 @@ DecodingOptions = {"task": ("ASR", "AED", "SER"),"language": "auto","fp16": True,+ "gain_event": True,}res = model.generate(input=input_wav, batch_size_s=0, DecodingOptions=DecodingOptions)diff --git a/funasr/models/paraformer/model.py b/funasr/models/paraformer/model.pyindex bd85df0d8..d47db1130 100644--- a/funasr/models/paraformer/model.py+++ b/funasr/models/paraformer/model.py@@ -475,6 +475,8 @@ class Paraformer(torch.nn.Module):speech = speech.to(device=kwargs["device"])speech_lengths = speech_lengths.to(device=kwargs["device"])# Encoder+ if kwargs.get("fp16", False):+ speech = speech.half()encoder_out, encoder_out_lens = self.encode(speech, speech_lengths)if isinstance(encoder_out, tuple):encoder_out = encoder_out[0]diff --git a/funasr/version.txt b/funasr/version.txtindex 2fa390179..1c2de38aa 100644--- a/funasr/version.txt+++ b/funasr/version.txt@@ -1 +1 @@-1.0.22\ No newline at end of file+1.0.23\ No newline at end of file