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

Browse Source
This commit is contained in:
嘉渊 2023-04-25 15:01:18 +08:00
parent 7f7c23c36f
commit 0237e32625
2 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
egs/aishell/paraformer/run.sh
View File

@ -3,8 +3,8 @@
. ./path.sh || exit 1; . ./path.sh || exit 1;
# machines configuration # machines configuration
CUDA_VISIBLE_DEVICES="0" CUDA_VISIBLE_DEVICES="0,1"
gpu_num=1 gpu_num=2
count=1 count=1
gpu_inference=true # Whether to perform gpu decoding, set false for cpu decoding gpu_inference=true # Whether to perform gpu decoding, set false for cpu decoding
# for gpu decoding, inference_nj=ngpu*njob; for cpu decoding, inference_nj=njob # for gpu decoding, inference_nj=ngpu*njob; for cpu decoding, inference_nj=njob

5
funasr/datasets/small_datasets/dataset.py
View File

@ -111,6 +111,7 @@ class ESPnetDataset(Dataset):
int_dtype: str = "long", int_dtype: str = "long",
dest_sample_rate: int = 16000, dest_sample_rate: int = 16000,
speed_perturb: tuple = None, speed_perturb: tuple = None,
mode: str = "train",
): ):
assert check_argument_types() assert check_argument_types()
if len(path_name_type_list) == 0: if len(path_name_type_list) == 0:
@ -125,6 +126,7 @@ class ESPnetDataset(Dataset):
self.int_dtype = int_dtype self.int_dtype = int_dtype
self.dest_sample_rate = dest_sample_rate self.dest_sample_rate = dest_sample_rate
self.speed_perturb = speed_perturb self.speed_perturb = speed_perturb
self.mode = mode
self.loader_dict = {} self.loader_dict = {}
self.debug_info = {} self.debug_info = {}
@ -148,7 +150,8 @@ class ESPnetDataset(Dataset):
loader_type: loader_type. sound, npy, text, etc loader_type: loader_type. sound, npy, text, etc
""" """
if loader_type == "sound": if loader_type == "sound":
loader = SoundScpReader(path, self.dest_sample_rate, normalize=True, always_2d=False, speed_perturb=self.speed_perturb) speed_perturb = self.speed_perturb if self.mode == "train" else None
loader = SoundScpReader(path, self.dest_sample_rate, normalize=True, always_2d=False, speed_perturb=speed_perturb)
return AdapterForSoundScpReader(loader, self.float_dtype) return AdapterForSoundScpReader(loader, self.float_dtype)
elif loader_type == "kaldi_ark": elif loader_type == "kaldi_ark":
loader = kaldiio.load_scp(path) loader = kaldiio.load_scp(path)