mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-09-19 05:54:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

disable speed perturbation by default

Browse Source
This commit is contained in:
zr_jin 2023-07-20 16:53:04 +08:00
parent 4ab7d61008
commit b8a34ae160
6 changed files with 100 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
egs/aidatatang_200zh/ASR/local/compute_fbank_aidatatang_200zh.py
View File

@ -42,7 +42,7 @@ torch.set_num_threads(1)
torch.set_num_interop_threads(1)
def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80, speed_perturb: bool = False):
src_dir = Path("data/manifests/aidatatang_200zh")
output_dir = Path("data/fbank")
num_jobs = min(15, os.cpu_count())
@ -86,8 +86,11 @@ def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
supervisions=m["supervisions"],
)
if "train" in partition:
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
cut_set
+ cut_set.perturb_speed(0.9)
+ cut_set.perturb_speed(1.1)
)
cut_set = cut_set.compute_and_store_features(
extractor=extractor,
@ -109,7 +112,12 @@ def get_args():
default=80,
help="""The number of mel bins for Fbank""",
)
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
@ -119,4 +127,6 @@ if __name__ == "__main__":
logging.basicConfig(format=formatter, level=logging.INFO)
args = get_args()
compute_fbank_aidatatang_200zh(num_mel_bins=args.num_mel_bins)
compute_fbank_aidatatang_200zh(
num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
)

18
egs/aishell/ASR/local/compute_fbank_aishell.py
View File

@ -42,7 +42,7 @@ torch.set_num_threads(1)
torch.set_num_interop_threads(1)
def compute_fbank_aishell(num_mel_bins: int = 80):
def compute_fbank_aishell(num_mel_bins: int = 80, speed_perturb: bool = False):
src_dir = Path("data/manifests")
output_dir = Path("data/fbank")
num_jobs = min(15, os.cpu_count())
@ -82,8 +82,11 @@ def compute_fbank_aishell(num_mel_bins: int = 80):
supervisions=m["supervisions"],
)
if "train" in partition:
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
cut_set
+ cut_set.perturb_speed(0.9)
+ cut_set.perturb_speed(1.1)
)
cut_set = cut_set.compute_and_store_features(
extractor=extractor,
@ -104,7 +107,12 @@ def get_args():
default=80,
help="""The number of mel bins for Fbank""",
)
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
@ -114,4 +122,6 @@ if __name__ == "__main__":
logging.basicConfig(format=formatter, level=logging.INFO)
args = get_args()
compute_fbank_aishell(num_mel_bins=args.num_mel_bins)
compute_fbank_aishell(
num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
)

17
egs/aishell2/ASR/local/compute_fbank_aishell2.py
View File

@ -42,7 +42,7 @@ torch.set_num_threads(1)
torch.set_num_interop_threads(1)
def compute_fbank_aishell2(num_mel_bins: int = 80):
def compute_fbank_aishell2(num_mel_bins: int = 80, speed_perturb: bool = False):
src_dir = Path("data/manifests")
output_dir = Path("data/fbank")
num_jobs = min(15, os.cpu_count())
@ -82,8 +82,11 @@ def compute_fbank_aishell2(num_mel_bins: int = 80):
supervisions=m["supervisions"],
)
if "train" in partition:
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
cut_set
+ cut_set.perturb_speed(0.9)
+ cut_set.perturb_speed(1.1)
)
cut_set = cut_set.compute_and_store_features(
extractor=extractor,
@ -104,6 +107,12 @@ def get_args():
default=80,
help="""The number of mel bins for Fbank""",
)
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
@ -114,4 +123,6 @@ if __name__ == "__main__":
logging.basicConfig(format=formatter, level=logging.INFO)
args = get_args()
compute_fbank_aishell2(num_mel_bins=args.num_mel_bins)
compute_fbank_aishell2(
num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
)

18
egs/aishell4/ASR/local/compute_fbank_aishell4.py
View File

@ -42,7 +42,7 @@ torch.set_num_threads(1)
torch.set_num_interop_threads(1)
def compute_fbank_aishell4(num_mel_bins: int = 80):
def compute_fbank_aishell4(num_mel_bins: int = 80, speed_perturb: bool = False):
src_dir = Path("data/manifests/aishell4")
output_dir = Path("data/fbank")
num_jobs = min(15, os.cpu_count())
@ -84,9 +84,13 @@ def compute_fbank_aishell4(num_mel_bins: int = 80):
supervisions=m["supervisions"],
)
if "train" in partition:
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
cut_set
+ cut_set.perturb_speed(0.9)
+ cut_set.perturb_speed(1.1)
)
cut_set = cut_set.compute_and_store_features(
extractor=extractor,
storage_path=f"{output_dir}/{prefix}_feats_{partition}",
@ -113,6 +117,12 @@ def get_args():
default=80,
help="""The number of mel bins for Fbank""",
)
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
@ -123,4 +133,6 @@ if __name__ == "__main__":
logging.basicConfig(format=formatter, level=logging.INFO)
args = get_args()
compute_fbank_aishell4(num_mel_bins=args.num_mel_bins)
compute_fbank_aishell4(
num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
)

17
egs/alimeeting/ASR/local/compute_fbank_alimeeting.py
View File

@ -42,7 +42,7 @@ torch.set_num_threads(1)
torch.set_num_interop_threads(1)
def compute_fbank_alimeeting(num_mel_bins: int = 80):
def compute_fbank_alimeeting(num_mel_bins: int = 80, speed_perturb: bool = False):
src_dir = Path("data/manifests/alimeeting")
output_dir = Path("data/fbank")
num_jobs = min(15, os.cpu_count())
@ -83,8 +83,11 @@ def compute_fbank_alimeeting(num_mel_bins: int = 80):
supervisions=m["supervisions"],
)
if "train" in partition:
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
cut_set
+ cut_set.perturb_speed(0.9)
+ cut_set.perturb_speed(1.1)
)
cur_num_jobs = num_jobs if ex is None else 80
cur_num_jobs = min(cur_num_jobs, len(cut_set))
@ -114,6 +117,12 @@ def get_args():
default=80,
help="""The number of mel bins for Fbank""",
)
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
@ -124,4 +133,6 @@ if __name__ == "__main__":
logging.basicConfig(format=formatter, level=logging.INFO)
args = get_args()
compute_fbank_alimeeting(num_mel_bins=args.num_mel_bins)
compute_fbank_alimeeting(
num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
)

22
egs/wenetspeech/ASR/local/preprocess_wenetspeech.py
View File

@ -16,6 +16,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import logging
import re
from pathlib import Path
@ -45,7 +46,7 @@ def has_no_oov(
return oov_pattern.search(sup.text) is None
def preprocess_wenet_speech():
def preprocess_wenet_speech(speed_perturb: bool = False):
src_dir = Path("data/manifests")
output_dir = Path("data/fbank")
output_dir.mkdir(exist_ok=True)
@ -115,15 +116,30 @@ def preprocess_wenet_speech():
f"Speed perturb for {partition} with factors 0.9 and 1.1 "
"(Perturbing may take 8 minutes and saving may take 20 minutes)"
)
cut_set = cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
if speed_perturb:
cut_set = (
cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
)
logging.info(f"Saving to {raw_cuts_path}")
cut_set.to_file(raw_cuts_path)
def get_args():
parser = argparse.ArgumentParser()
parser.add_argument(
"--speed-perturb",
type=bool,
default=False,
help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
)
return parser.parse_args()
def main():
setup_logger(log_filename="./log-preprocess-wenetspeech")
preprocess_wenet_speech()
args = get_args()
preprocess_wenet_speech(speed_perturb=args.speed_perturb)
logging.info("Done")