fixed bug of compute_cmvn_stats.py

examples/hi_xiaowen/s0/conf/ds_tcn.yaml

@@ -5,11 +5,12 @@ dataset_conf:
     resample_conf:
         resample_rate: 16000
     speed_perturb: false
-    fbank_conf:
+    feature_extraction_conf:
+        feature_type: 'fbank'
         num_mel_bins: 40
         frame_shift: 10
         frame_length: 25
-        dither: 0.1
+        dither: 1.0
     spec_aug: true
     spec_aug_conf:
         num_t_mask: 1
@@ -24,9 +25,9 @@ dataset_conf:
 
 model:
     hidden_dim: 64
-    subsampling:
+    preprocessing:
         type: linear
-    body:
+    backbone:
         type: tcn
         ds: true
         num_layers: 4

examples/hi_xiaowen/s0/conf/gru.yaml

@@ -5,11 +5,12 @@ dataset_conf:
     resample_conf:
         resample_rate: 16000
     speed_perturb: false
-    fbank_conf:
+    feature_extraction_conf:
+        feature_type: 'fbank'
         num_mel_bins: 40
         frame_shift: 10
         frame_length: 25
-        dither: 0.1
+        dither: 1.0
     spec_aug: false
     spec_aug_conf:
         num_t_mask: 2
@@ -24,9 +25,9 @@ dataset_conf:
 
 model:
     hidden_dim: 128
-    subsampling:
+    preprocessing:
         type: linear
-    body:
+    backbone:
         type: gru
         num_layers: 2
 

examples/hi_xiaowen/s0/conf/mdtc.yaml

@@ -1,7 +1,4 @@
-debug: false
 
-input_dim: 80
-output_dim: 2
 
 dataset_conf:
     filter_conf:
@@ -38,6 +35,8 @@ dataset_conf:
         batch_size: 100
 
 model:
+    input_dim: 80
+    output_dim: 2
     hidden_dim: 64
     preprocessing:
         type: none

examples/hi_xiaowen/s0/conf/tcn.yaml

@@ -5,11 +5,12 @@ dataset_conf:
     resample_conf:
         resample_rate: 16000
     speed_perturb: false
-    fbank_conf:
+    feature_extraction_conf:
+        feature_type: 'fbank'
         num_mel_bins: 40
         frame_shift: 10
         frame_length: 25
-        dither: 0.1
+        dither: 1.0
     spec_aug: false
     spec_aug_conf:
         num_t_mask: 2
@@ -24,9 +25,9 @@ dataset_conf:
 
 model:
     hidden_dim: 64
-    subsampling:
+    preprocessing:
         type: linear
-    body:
+    backbone:
         type: tcn
         ds: false
         num_layers: 4

examples/hi_xiaowen/s0/run.sh

@@ -5,8 +5,8 @@
 
 export CUDA_VISIBLE_DEVICES="0"
 
-stage=2
+stage=0
-stop_stage=2
+stop_stage=4
 num_keywords=2
 
 config=conf/mdtc.yaml

kws/bin/train.py

@@ -135,7 +135,7 @@ def main():
                                 num_workers=args.num_workers,
                                 prefetch_factor=args.prefetch)
 
-    input_dim = configs['input_dim']
+    input_dim = configs['dataset_conf']['feature_extraction_conf']['num_mel_bins']
     output_dim = args.num_keywords
 
     # Write model_dir/config.yaml for inference and export

kws/dataset/processor.py

@@ -165,6 +165,7 @@ def compute_mfcc(
         yield dict(key=sample['key'], label=sample['label'], feat=mat)
 
 def compute_fbank(data,
+                  feature_type='fbank',
                   num_mel_bins=23,
                   frame_length=25,
                   frame_shift=10,

tools/compute_cmvn_stats.py

@@ -18,9 +18,10 @@ torchaudio.set_audio_backend("sox_io")
 class CollateFunc(object):
     ''' Collate function for AudioDataset
     '''
-    def __init__(self, feat_dim, resample_rate):
+    def __init__(self, feat_dim, feat_type, resample_rate):
         self.feat_dim = feat_dim
         self.resample_rate = resample_rate
+        self.feat_type = feat_type
         pass
 
     def __call__(self, batch):
@@ -50,12 +51,20 @@ class CollateFunc(object):
                 resample_rate = self.resample_rate
                 waveform = torchaudio.transforms.Resample(
                     orig_freq=sample_rate, new_freq=resample_rate)(waveform)
-
+            if self.feat_type == 'fbank':
                 mat = kaldi.fbank(waveform,
                               num_mel_bins=self.feat_dim,
                               dither=0.0,
                               energy_floor=0.0,
                               sample_frequency=resample_rate)
+            elif self.feat_type == 'mfcc':
+                mat = kaldi.mfcc(waveform,
+                        num_ceps=self.feat_dim,
+                        num_mel_bins=self.feat_dim,
+                        dither=0.0,
+                        energy_floor=0.0,
+                        sample_frequency=resample_rate,
+        )
             mean_stat += torch.sum(mat, axis=0)
             var_stat += torch.sum(torch.square(mat), axis=0)
             number += mat.shape[0]
@@ -95,13 +104,14 @@ if __name__ == '__main__':
 
     with open(args.train_config, 'r') as fin:
         configs = yaml.load(fin, Loader=yaml.FullLoader)
-    feat_dim = configs['input_dim']
+    feat_dim = configs['dataset_conf']['feature_extraction_conf']['num_mel_bins']
+    feat_type = configs['dataset_conf']['feature_extraction_conf']['feature_type']
     resample_rate = 0
     if 'resample_conf' in configs['dataset_conf']:
         resample_rate = configs['dataset_conf']['resample_conf']['resample_rate']
         print('using resample and new sample rate is {}'.format(resample_rate))
 
-    collate_func = CollateFunc(feat_dim, resample_rate)
+    collate_func = CollateFunc(feat_dim, feat_type, resample_rate)
     dataset = AudioDataset(args.in_scp)
     batch_size = 20
     data_loader = DataLoader(dataset,