update

examples/vm_sound_classification/run.sh

@@ -15,7 +15,7 @@ E:/Users/tianx/HuggingDatasets/vm_sound_classification/data/wav_finished/id-ID/w
 sh run.sh --stage 2 --stop_stage 2 --system_version windows --file_folder_name file_dir --final_model_name vm_sound_classification3 --label_plan 4
 sh run.sh --stage 3 --stop_stage 3 --system_version windows --file_folder_name file_dir --final_model_name vm_sound_classification3 --label_plan 4
 
-sh run.sh --stage 2 --stop_stage 5 --system_version centos --file_folder_name file_dir --final_model_name vm_sound_classification4-ch16 \
+sh run.sh --stage 4 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name vm_sound_classification4-ch16 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" --label_plan 4
 
 "

toolbox/torchaudio/augment/spec_augment.py

@@ -8,6 +8,7 @@ from typing import List, Tuple
 
 import torch
 import torch.nn as nn
+from torch.distributions import uniform
 
 
 class SpecAugment(nn.Module):
@@ -19,8 +20,8 @@ class SpecAugment(nn.Module):
 
     @staticmethod
     def augment_volume(spec: torch.Tensor, factor_range: Tuple[float, float] = (0.5, 2.0)):
-        factor = random.uniform(*factor_range)
-        factor = torch.tensor(factor, dtype=torch.float32)
+        factor = uniform.Uniform(*factor_range)
+        factor = factor.sample()
         spec_ = spec.clone().detach()
         spec_ *= factor
         return spec_