speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

update (debugging)

Browse Source
This commit is contained in:
shixian.shi 2024-01-12 17:59:15 +08:00
parent 38524b21d9
commit bcb8b0c3cb
2 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
funasr/bin/inference.py
View File

@ -115,7 +115,7 @@ class AutoModel:
vad_model = kwargs.get("vad_model", None)
vad_kwargs = kwargs.get("vad_model_revision", None)
if vad_model is not None:
print("build vad model")
logging.info("Building VAD model.")
vad_kwargs = {"model": vad_model, "model_revision": vad_kwargs}
vad_model, vad_kwargs = self.build_model(**vad_kwargs)
@ -123,6 +123,7 @@ class AutoModel:
punc_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}
punc_model, punc_kwargs = self.build_model(**punc_kwargs)
@ -130,6 +131,7 @@ class AutoModel:
spk_model = kwargs.get("spk_model", None)
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}
spk_model, spk_kwargs = self.build_model(**spk_kwargs)
self.cb_model = ClusterBackend()
@ -166,6 +168,7 @@ class AutoModel:
device = "cpu"
# kwargs["batch_size"] = 1
kwargs["device"] = device
import pdb; pdb.set_trace()
if kwargs.get("ncpu", None):
torch.set_num_threads(kwargs.get("ncpu"))

3
funasr/models/seaco_paraformer/model.py
View File

@ -337,7 +337,8 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
meta_data[
"batch_data_time"] = speech_lengths.sum().item() * frontend.frame_shift * frontend.lfr_n / 1000
speech.to(device=kwargs["device"]), speech_lengths.to(device=kwargs["device"])
speech = speech.to(device=kwargs["device"])
speech_lengths = speech_lengths.to(device=kwargs["device"])
# hotword
self.hotword_list = self.generate_hotwords_list(kwargs.get("hotword", None), tokenizer=tokenizer, frontend=frontend)