LibriParty/VAD Bug

Describe the bug

cd recipes/LibriParty/VAD
python train.py hparams/train.yaml --data_folder=/path/to/LibriParty --musan_folder=/path/to/musan/ --commonlanguage_folder=/path/to/common_voice_kpd

Traceback (most recent call last):
File "/ai/phj/research/VAD/vad/VAD/train.py", line 278, in
vad_brain.fit(
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/core.py", line 1153, in fit
self._fit_train(train_set=train_set, epoch=epoch, enable=enable)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/core.py", line 1009, in _fit_train
loss = self.fit_batch(batch)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/core.py", line 881, in fit_batch
outputs = self.compute_forward(batch, Stage.TRAIN)
File "/ai/phj/research/VAD/vad/VAD/train.py", line 50, in compute_forward
wavs, targets, lens = augment_data(
File "/ai/phj/research/VAD/vad/VAD/data_augment.py", line 334, in augment_data
wav_samples_noise = get_samples_from_datasets(noise_datasets, wavs)
File "/ai/phj/research/VAD/vad/VAD/data_augment.py", line 186, in get_samples_from_datasets
wav_sample = dataset(wav_sample, len_sample)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/lobes/augment.py", line 446, in forward
waveforms = self.add_noise(waveforms, lengths)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/processing/speech_augmentation.py", line 158, in forward
noise_waveform, noise_length = self._load_noise(
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/processing/speech_augmentation.py", line 188, in _load_noise
dataset = ExtendedCSVDataset(
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/dataio/legacy.py", line 92, in init
super().init(data, dynamic_items, output_keys)
File "/ai/tool/miniconda3/envs/vad/lib/python3.9/site-packages/speechbrain/dataio/dataset.py", line 151, in init
static_keys = list(self.data[self.data_ids[0]].keys())
IndexError: list index out of range

Expected behaviour

The above is the error I encountered when running the VAD script. May I ask how to run it normally！

To Reproduce

No response

Versions

No response

Relevant log output

No response

Additional context

No response

Start by reproducing the command in recipes/LibriParty/VAD with the configured data paths, then trace the failure from train.py into data_augment.py, especially get_samples_from_datasets. Check how the noise dataset is constructed before SpeechBrain’s speech_augmentation.py receives it. Done means identifying the cause of the empty dataset and confirming that the VAD training command runs past this traceback.