mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-09 10:02:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

changes to train script - no need for limiting utterance length here

Browse Source
This commit is contained in:
Bailey Hirota 2025-06-13 00:48:37 +09:00
parent e1f140a50e
commit fe9f975ec2
2 changed files with 4 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
egs/multi_ja_en/ASR/local/validate_bpe_lexicon.py
View File

@ -1 +1 @@
/root/icefall/egs/librispeech/ASR/local/validate_bpe_lexicon.py /root/Github/reazon-icefall/egs/librispeech/ASR/local/validate_bpe_lexicon.py

14
egs/multi_ja_en/ASR/zipformer/train.py
View File

@ -1185,15 +1185,12 @@ def run(rank, world_size, args):
train_cuts = multi_dataset.train_cuts() train_cuts = multi_dataset.train_cuts()
def remove_short_and_long_utt(c: Cut): def remove_short_and_long_utt(c: Cut):
# Keep only utterances with duration between 1 second and 30 seconds # Keep only utterances greater than 1 second
#
# Caution: There is a reason to select 30.0 here. Please see
# ../local/display_manifest_statistics.py
# #
# You should use ../local/display_manifest_statistics.py to get # You should use ../local/display_manifest_statistics.py to get
# an utterance duration distribution for your dataset to select # an utterance duration distribution for your dataset to select
# the threshold # the threshold as this is dependent on which datasets you choose
if c.duration < 1.0 or c.duration > 30.0: if c.duration < 1.0:
logging.warning( logging.warning(
f"Exclude cut with ID {c.id} from training. Duration: {c.duration}" f"Exclude cut with ID {c.id} from training. Duration: {c.duration}"
) )
@ -1239,14 +1236,10 @@ def run(rank, world_size, args):
else: else:
sampler_state_dict = None sampler_state_dict = None
# train_dl = reazonspeech_corpus.train_dataloaders(
# train_cuts, sampler_state_dict=sampler_state_dict
# )
train_dl = multidataset_datamodule.train_dataloaders( train_dl = multidataset_datamodule.train_dataloaders(
train_cuts, sampler_state_dict=sampler_state_dict train_cuts, sampler_state_dict=sampler_state_dict
) )
valid_cuts = multi_dataset.dev_cuts() valid_cuts = multi_dataset.dev_cuts()
valid_dl = multidataset_datamodule.valid_dataloaders(valid_cuts) valid_dl = multidataset_datamodule.valid_dataloaders(valid_cuts)
@ -1393,7 +1386,6 @@ def main():
MultiDatasetAsrDataModule.add_arguments(parser) MultiDatasetAsrDataModule.add_arguments(parser)
args = parser.parse_args() args = parser.parse_args()
args.exp_dir = Path(args.exp_dir) args.exp_dir = Path(args.exp_dir)
print(args)
world_size = args.world_size world_size = args.world_size
assert world_size >= 1 assert world_size >= 1