mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-09 01:52:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix for asr_datamodule.py

Browse Source
This commit is contained in:
Fangjun Kuang 2025-07-01 17:20:27 +08:00
parent 85f6deb8d1
commit eaaab47509
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py
View File

@ -29,7 +29,6 @@ from lhotse.dataset import ( # noqa F401 for PrecomputedFeatures
CutConcatenate, CutConcatenate,
CutMix, CutMix,
DynamicBucketingSampler, DynamicBucketingSampler,
K2SpeechRecognitionDataset,
PrecomputedFeatures, PrecomputedFeatures,
SimpleCutSampler, SimpleCutSampler,
SpecAugment, SpecAugment,
@ -39,6 +38,7 @@ from lhotse.dataset.input_strategies import ( # noqa F401 For AudioSamples
OnTheFlyFeatures, OnTheFlyFeatures,
) )
from lhotse.utils import fix_random_seed from lhotse.utils import fix_random_seed
from speech_recognition import K2SpeechRecognitionDataset
from torch.utils.data import DataLoader from torch.utils.data import DataLoader
from icefall.utils import str2bool from icefall.utils import str2bool
@ -232,8 +232,11 @@ class LibriSpeechAsrDataModule:
logging.info("Enable MUSAN") logging.info("Enable MUSAN")
logging.info("About to get Musan cuts") logging.info("About to get Musan cuts")
cuts_musan = load_manifest(self.args.manifest_dir / "musan_cuts.jsonl.gz") cuts_musan = load_manifest(self.args.manifest_dir / "musan_cuts.jsonl.gz")
# We use probability 1.0 here so that musan augmentation is
# always performed
transforms.append( transforms.append(
CutMix(cuts=cuts_musan, p=0.5, snr=(10, 20), preserve_id=True) CutMix(cuts=cuts_musan, p=1.0, snr=(10, 20), preserve_id=True)
) )
else: else:
logging.info("Disable MUSAN") logging.info("Disable MUSAN")