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
vad conf
游雁fae73ee41441
diff --git a/examples/industrial_data_pretraining/paraformer/demo.py b/examples/industrial_data_pretraining/paraformer/demo.pyindex 965f8f3fc..499791fa2 100644--- a/examples/industrial_data_pretraining/paraformer/demo.py+++ b/examples/industrial_data_pretraining/paraformer/demo.py@@ -7,10 +7,11 @@ from funasr import AutoModelmodel = AutoModel(model="iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch",model_revision="v2.0.4",- # vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",- # vad_model_revision="v2.0.4",- # punc_model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch",- # punc_model_revision="v2.0.4",+ vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",+ vad_model_revision="v2.0.4",+ vad_kwargs={"max_single_segment_time": 60},+ punc_model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch",+ punc_model_revision="v2.0.4",# spk_model="iic/speech_campplus_sv_zh-cn_16k-common",# spk_model_revision="v2.0.2",)diff --git a/examples/industrial_data_pretraining/whisper/demo.py b/examples/industrial_data_pretraining/whisper/demo.pyindex ddebbdfe7..a073f68f3 100644--- a/examples/industrial_data_pretraining/whisper/demo.py+++ b/examples/industrial_data_pretraining/whisper/demo.py@@ -10,6 +10,7 @@ from funasr import AutoModelmodel = AutoModel(model="iic/Whisper-large-v3",model_revision="v2.0.5",vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",+ vad_kwargs={"max_single_segment_time": 30},)res = model.generate(diff --git a/examples/industrial_data_pretraining/whisper/demo_from_openai.py b/examples/industrial_data_pretraining/whisper/demo_from_openai.pyindex 5cac06b5d..9a2276445 100644--- a/examples/industrial_data_pretraining/whisper/demo_from_openai.py+++ b/examples/industrial_data_pretraining/whisper/demo_from_openai.py@@ -10,7 +10,11 @@ from funasr import AutoModel# model = AutoModel(model="Whisper-small", hub="openai")# model = AutoModel(model="Whisper-medium", hub="openai")# model = AutoModel(model="Whisper-large-v2", hub="openai")-model = AutoModel(model="Whisper-large-v3", hub="openai", vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",)+model = AutoModel(model="Whisper-large-v3",+ vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",+ vad_kwargs={"max_single_segment_time": 30},+ hub="openai",+ )res = model.generate(language=None,diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.pyindex 69aef28fc..39f91e977 100644--- a/funasr/auto/auto_model.py+++ b/funasr/auto/auto_model.py@@ -68,7 +68,8 @@ def prepare_data_iterator(data_in, input_len=None, data_type=None, key=None):data_list.append(data)key_list.append(key)else:- key = "rand_key_" + ''.join(random.choice(chars) for _ in range(13))+ if key is None:+ key = "rand_key_" + ''.join(random.choice(chars) for _ in range(13))data_list = [data_in]key_list = [key]elif isinstance(data_in, (list, tuple)):@@ -105,18 +106,23 @@ class AutoModel:# if vad_model is not None, build vad model else Nonevad_model = kwargs.get("vad_model", None)- vad_kwargs = kwargs.get("vad_model_revision", None)if vad_model is not None:logging.info("Building VAD model.")- vad_kwargs = {"model": vad_model, "model_revision": vad_kwargs, "device": kwargs["device"]}+ vad_kwargs = {} if kwargs.get("vad_kwargs", {}) is None else kwargs.get("vad_kwargs", {})+ vad_kwargs["model"] = vad_model+ vad_kwargs["model_revision"] = kwargs.get("vad_model_revision", None)+ vad_kwargs["device"] = kwargs["device"]vad_model, vad_kwargs = self.build_model(**vad_kwargs)# if punc_model is not None, build punc model else Nonepunc_model = kwargs.get("punc_model", None)- punc_kwargs = kwargs.get("punc_model_revision", None)+if punc_model is not None:logging.info("Building punc model.")- punc_kwargs = {"model": punc_model, "model_revision": punc_kwargs, "device": kwargs["device"]}+ punc_kwargs = {} if kwargs.get("punc_kwargs", {}) is None else kwargs.get("punc_kwargs", {})+ punc_kwargs["model"] = punc_model+ punc_kwargs["model_revision"] = kwargs.get("punc_model_revision", None)+ punc_kwargs["device"] = kwargs["device"]punc_model, punc_kwargs = self.build_model(**punc_kwargs)# if spk_model is not None, build spk model else None@@ -124,7 +130,10 @@ class AutoModel:spk_kwargs = kwargs.get("spk_model_revision", None)if spk_model is not None:logging.info("Building SPK model.")- spk_kwargs = {"model": spk_model, "model_revision": spk_kwargs, "device": kwargs["device"]}+ spk_kwargs = {} if kwargs.get("spk_kwargs", {}) is None else kwargs.get("spk_kwargs", {})+ spk_kwargs["model"] = spk_model+ spk_kwargs["model_revision"] = kwargs.get("spk_model_revision", None)+ spk_kwargs["device"] = kwargs["device"]spk_model, spk_kwargs = self.build_model(**spk_kwargs)self.cb_model = ClusterBackend().to(kwargs["device"])spk_mode = kwargs.get("spk_mode", 'punc_segment')@@ -162,10 +171,7 @@ class AutoModel:tokenizer = kwargs.get("tokenizer", None)if tokenizer is not None:tokenizer_class = tables.tokenizer_classes.get(tokenizer)- tokenizer_conf = kwargs.get("tokenizer_conf", {})- tokenizer = tokenizer_class(**tokenizer_conf)--+ tokenizer = tokenizer_class(**kwargs.get("tokenizer_conf", {}))kwargs["token_list"] = tokenizer.token_list if hasattr(tokenizer, "token_list") else Nonekwargs["token_list"] = tokenizer.get_vocab() if hasattr(tokenizer, "get_vocab") else kwargs["token_list"]vocab_size = len(kwargs["token_list"]) if kwargs["token_list"] is not None else -1@@ -178,12 +184,14 @@ class AutoModel:kwargs["input_size"] = Noneif frontend is not None:frontend_class = tables.frontend_classes.get(frontend)- frontend = frontend_class(**kwargs["frontend_conf"])+ frontend = frontend_class(**kwargs.get("frontend_conf", {}))kwargs["input_size"] = frontend.output_size() if hasattr(frontend, "output_size") else Nonekwargs["frontend"] = frontend# build modelmodel_class = tables.model_classes.get(kwargs["model"])- model = model_class(**kwargs, **kwargs.get("model_conf", {}), vocab_size=vocab_size)+ model_conf = kwargs.get("model_conf", {})+ deep_update(model_conf, kwargs)+ model = model_class(**model_conf, vocab_size=vocab_size)model.to(device)# init_param