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
test
语帆491b2af1ecdf
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.pyindex 9db8c015d..68559d121 100644--- a/funasr/auto/auto_model.py+++ b/funasr/auto/auto_model.py@@ -153,12 +153,9 @@ class AutoModel:# build tokenizertokenizer = kwargs.get("tokenizer", None)- pdb.set_trace()if tokenizer is not None:tokenizer_class = tables.tokenizer_classes.get(tokenizer)- pdb.set_trace()tokenizer = tokenizer_class(**kwargs["tokenizer_conf"])- pdb.set_trace()kwargs["tokenizer"] = tokenizerkwargs["token_list"] = tokenizer.token_listvocab_size = len(tokenizer.token_list)diff --git a/funasr/frontends/default.py b/funasr/frontends/default.pyindex 8ac1ca853..15cc35a27 100644--- a/funasr/frontends/default.py+++ b/funasr/frontends/default.py@@ -17,7 +17,7 @@ from funasr.frontends.utils.stft import Stftfrom funasr.frontends.utils.frontend import Frontendfrom funasr.models.transformer.utils.nets_utils import make_pad_mask-[email protected]("frontend_classes", "DefaultFrontend")class DefaultFrontend(nn.Module):"""Conventional frontend structure for ASR.Stft -> WPE -> MVDR-Beamformer -> Power-spec -> Mel-Fbank -> CMVN