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

test

Browse Source
This commit is contained in:
语帆 2024-02-28 13:56:32 +08:00
parent 5ecd13bd04
commit 343a281ca1
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
funasr/models/lcbnet/model.py
View File

@ -413,7 +413,6 @@ class LCBNet(nn.Module):
logging.info("enable beam_search")
self.init_beam_search(**kwargs)
self.nbest = kwargs.get("nbest", 1)
pdb.set_trace()
meta_data = {}
if isinstance(data_in, torch.Tensor) and kwargs.get("data_type", "sound") == "fbank": # fbank
@ -431,6 +430,7 @@ class LCBNet(nn.Module):
tokenizer=tokenizer)
time2 = time.perf_counter()
meta_data["load_data"] = f"{time2 - time1:0.3f}"
pdb.set_trace()
speech, speech_lengths = extract_fbank(audio_sample_list, data_type=kwargs.get("data_type", "sound"),
frontend=frontend)
time3 = time.perf_counter()

6
funasr/utils/load_utils.py
View File

@ -31,14 +31,13 @@ def load_audio_text_image_video(data_or_path_or_list, fs: int = 16000, audio_fs:
return [load_audio_text_image_video(audio, fs=fs, audio_fs=audio_fs, data_type=data_type, **kwargs) for audio in data_or_path_or_list]
if isinstance(data_or_path_or_list, str) and data_or_path_or_list.startswith('http'): # download url to local file
data_or_path_or_list = download_from_url(data_or_path_or_list)
pdb.set_trace()
if isinstance(data_or_path_or_list, str) and os.path.exists(data_or_path_or_list): # local file
if data_type is None or data_type == "sound":
data_or_path_or_list, audio_fs = torchaudio.load(data_or_path_or_list)
if kwargs.get("reduce_channels", True):
data_or_path_or_list = data_or_path_or_list.mean(0)
elif data_type == "text" and tokenizer is not None:
pdb.set_trace()
data_or_path_or_list = tokenizer.encode(data_or_path_or_list)
elif data_type == "image": # undo
pass
@ -68,7 +67,7 @@ def load_audio_text_image_video(data_or_path_or_list, fs: int = 16000, audio_fs:
else:
pass
# print(f"unsupport data type: {data_or_path_or_list}, return raw data")
pdb.set_trace()
if audio_fs != fs and data_type != "text":
resampler = torchaudio.transforms.Resample(audio_fs, fs)
data_or_path_or_list = resampler(data_or_path_or_list[None, :])[0, :]
@ -112,6 +111,7 @@ def extract_fbank(data, data_len = None, data_type: str="sound", frontend=None,
# import pdb;
# pdb.set_trace()
# if data_type == "sound":
pdb.set_trace()
data, data_len = frontend(data, data_len, **kwargs)
if isinstance(data_len, (list, tuple)):