mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-25 17:56:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

implement musan and enable by default

Browse Source
This commit is contained in:
Kinan Martin 2025-03-10 10:12:45 +09:00
parent de7406f1b9
commit c5c35859ec
2 changed files with 21 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
egs/reazonspeech/ASR/local/utils/asr_datamodule.py
View File

@ -174,13 +174,16 @@ class ReazonSpeechAsrDataModule:
group.add_argument( group.add_argument(
"--enable-musan", "--enable-musan",
type=str2bool, type=str2bool,
default=False, default=True,
help="When enabled, select noise from MUSAN and mix it" help="When enabled, select noise from MUSAN and mix it"
"with training dataset. ", "with training dataset. ",
) )
def train_dataloaders( def train_dataloaders(
self, cuts_train: CutSet, sampler_state_dict: Optional[Dict[str, Any]] = None self,
cuts_train: CutSet,
sampler_state_dict: Optional[Dict[str, Any]] = None,
cuts_musan: Optional[CutSet] = None,
) -> DataLoader: ) -> DataLoader:
""" """
Args: Args:
@ -191,6 +194,14 @@ class ReazonSpeechAsrDataModule:
""" """
transforms = [] transforms = []
if cuts_musan is not None:
logging.info("Enable MUSAN")
transforms.append(
CutMix(cuts=cuts_musan, p=0.5, snr=(10, 20), preserve_id=True)
)
else:
logging.info("Disable MUSAN")
input_transforms = [] input_transforms = []
if self.args.enable_spec_aug: if self.args.enable_spec_aug:

9
egs/reazonspeech/ASR/zipformer/train.py
View File

@ -1219,8 +1219,15 @@ def run(rank, world_size, args):
else: else:
sampler_state_dict = None sampler_state_dict = None
if args.enable_musan:
cuts_musan = load_manifest(Path(args.manifest_dir) / "musan_cuts.jsonl.gz")
else:
cuts_musan = None
train_dl = reazonspeech_corpus.train_dataloaders( train_dl = reazonspeech_corpus.train_dataloaders(
train_cuts, sampler_state_dict=sampler_state_dict train_cuts,
sampler_state_dict=sampler_state_dict,
cuts_musan=cuts_musan,
) )
valid_cuts = reazonspeech_corpus.valid_cuts() valid_cuts = reazonspeech_corpus.valid_cuts()