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

load_audio_text_image_video

Browse Source
This commit is contained in:
游雁 2024-01-05 17:00:11 +08:00
parent 4f98546f36
commit e6a7bbe1ca
1 changed files with 29 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
funasr/utils/load_utils.py
View File

@ -10,52 +10,52 @@ import time
import logging import logging
from torch.nn.utils.rnn import pad_sequence from torch.nn.utils.rnn import pad_sequence
# def load_audio(audio_or_path_or_list, fs: int=16000, audio_fs: int=16000): # def load_audio(data_or_path_or_list, fs: int=16000, audio_fs: int=16000):
# #
# if isinstance(audio_or_path_or_list, (list, tuple)): # if isinstance(data_or_path_or_list, (list, tuple)):
# return [load_audio(audio, fs=fs, audio_fs=audio_fs) for audio in audio_or_path_or_list] # return [load_audio(audio, fs=fs, audio_fs=audio_fs) for audio in data_or_path_or_list]
# #
# if isinstance(audio_or_path_or_list, str) and os.path.exists(audio_or_path_or_list): # if isinstance(data_or_path_or_list, str) and os.path.exists(data_or_path_or_list):
# audio_or_path_or_list, audio_fs = torchaudio.load(audio_or_path_or_list) # data_or_path_or_list, audio_fs = torchaudio.load(data_or_path_or_list)
# audio_or_path_or_list = audio_or_path_or_list[0, :] # data_or_path_or_list = data_or_path_or_list[0, :]
# elif isinstance(audio_or_path_or_list, np.ndarray): # audio sample point # elif isinstance(data_or_path_or_list, np.ndarray): # audio sample point
# audio_or_path_or_list = np.squeeze(audio_or_path_or_list) #[n_samples,] # data_or_path_or_list = np.squeeze(data_or_path_or_list) #[n_samples,]
# #
# if audio_fs != fs: # if audio_fs != fs:
# resampler = torchaudio.transforms.Resample(audio_fs, fs) # resampler = torchaudio.transforms.Resample(audio_fs, fs)
# audio_or_path_or_list = resampler(audio_or_path_or_list[None, :])[0, :] # data_or_path_or_list = resampler(data_or_path_or_list[None, :])[0, :]
# return audio_or_path_or_list # return data_or_path_or_list
def load_audio_text_image_video(audio_or_path_or_list, fs: int = 16000, audio_fs: int = 16000, data_type=None, tokenizer=None): def load_audio_text_image_video(data_or_path_or_list, fs: int = 16000, audio_fs: int = 16000, data_type=None, tokenizer=None):
if isinstance(audio_or_path_or_list, (list, tuple)): if isinstance(data_or_path_or_list, (list, tuple)):
if data_type is not None and isinstance(data_type, (list, tuple)): if data_type is not None and isinstance(data_type, (list, tuple)):
data_types = [data_type] * len(audio_or_path_or_list) data_types = [data_type] * len(data_or_path_or_list)
audio_or_path_or_list_ret = [[] for d in data_type] data_or_path_or_list_ret = [[] for d in data_type]
for i, (data_type_i, audio_or_path_or_list_i) in enumerate(zip(data_types, audio_or_path_or_list)): for i, (data_type_i, data_or_path_or_list_i) in enumerate(zip(data_types, data_or_path_or_list)):
for j, (data_type_j, audio_or_path_or_list_j) in enumerate(zip(data_type_i, audio_or_path_or_list_i)): for j, (data_type_j, data_or_path_or_list_j) in enumerate(zip(data_type_i, data_or_path_or_list_i)):
audio_or_path_or_list_j = load_audio_text_image_video(audio_or_path_or_list_j, fs=fs, audio_fs=audio_fs, data_type=data_type_j, tokenizer=tokenizer) data_or_path_or_list_j = load_audio_text_image_video(data_or_path_or_list_j, fs=fs, audio_fs=audio_fs, data_type=data_type_j, tokenizer=tokenizer)
audio_or_path_or_list_ret[j].append(audio_or_path_or_list_j) data_or_path_or_list_ret[j].append(data_or_path_or_list_j)
return audio_or_path_or_list_ret return data_or_path_or_list_ret
else: else:
return [load_audio_text_image_video(audio, fs=fs, audio_fs=audio_fs) for audio in audio_or_path_or_list] return [load_audio_text_image_video(audio, fs=fs, audio_fs=audio_fs) for audio in data_or_path_or_list]
if isinstance(audio_or_path_or_list, str) and os.path.exists(audio_or_path_or_list): if isinstance(data_or_path_or_list, str) and os.path.exists(data_or_path_or_list):
audio_or_path_or_list, audio_fs = torchaudio.load(audio_or_path_or_list) data_or_path_or_list, audio_fs = torchaudio.load(data_or_path_or_list)
audio_or_path_or_list = audio_or_path_or_list[0, :] data_or_path_or_list = data_or_path_or_list[0, :]
elif isinstance(audio_or_path_or_list, np.ndarray): # audio sample point elif isinstance(data_or_path_or_list, np.ndarray): # audio sample point
audio_or_path_or_list = np.squeeze(audio_or_path_or_list) # [n_samples,] data_or_path_or_list = np.squeeze(data_or_path_or_list) # [n_samples,]
elif isinstance(audio_or_path_or_list, str) and data_type is not None and data_type == "text" and tokenizer is not None: elif isinstance(data_or_path_or_list, str) and data_type is not None and data_type == "text" and tokenizer is not None:
audio_or_path_or_list = tokenizer.encode(audio_or_path_or_list) data_or_path_or_list = tokenizer.encode(data_or_path_or_list)
if audio_fs != fs and data_type != "text": if audio_fs != fs and data_type != "text":
resampler = torchaudio.transforms.Resample(audio_fs, fs) resampler = torchaudio.transforms.Resample(audio_fs, fs)
audio_or_path_or_list = resampler(audio_or_path_or_list[None, :])[0, :] data_or_path_or_list = resampler(data_or_path_or_list[None, :])[0, :]
return audio_or_path_or_list return data_or_path_or_list
def load_bytes(input): def load_bytes(input):
middle_data = np.frombuffer(input, dtype=np.int16) middle_data = np.frombuffer(input, dtype=np.int16)