Hugging Face's logo

Lingeshg
/
SpeechEmotionDetector

Voice Activity Detection
Model card Files
xet
Community
Lingeshg commited on
Commit
fe65b7d
·
verified ·
1 Parent(s): 809d399

Upload 8 files

Browse files
Files changed (8) hide show
  1. hparams/train.yaml +35 -0
  2. hparams/train_with_wav2vec.yaml +112 -0
  3. model.pth +3 -0
  4. prepare.py +52 -0
  5. results/train_with_wav2vec2/1993/test.json +1 -0
  6. results/train_with_wav2vec2/1993/train.json +1 -0
  7. results/train_with_wav2vec2/1993/valid.json +1 -0
  8. train_with_wav2vec.py +302 -0
hparams/train.yaml ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ seed: 1993
2
+ __set_seed: !apply:torch.manual_seed [!ref <seed>]
3
+
4
+ # Dataset will be downloaded to the `data_original`
5
+ data_original: D:/voice-emo/dat/
6
+ output_folder: results/train_with_wav2vec2/1993
7
+ save_folder: !ref <output_folder>/save
8
+ train_log: !ref <output_folder>/train_log.txt
9
+
10
+ # URL for the wav2vec2 model
11
+ wav2vec2_hub: facebook/wav2vec2-base
12
+ wav2vec2_folder: !ref <save_folder>/wav2vec2_checkpoint
13
+
14
+ # Path where data manifest files will be stored
15
+ train_annotation: !ref <output_folder>/train.json
16
+ valid_annotation: !ref <output_folder>/valid.json
17
+ test_annotation: !ref <output_folder>/test.json
18
+ split_ratio: [80, 10, 10]
19
+ skip_prep: False
20
+
21
+ number_of_epochs: 5
22
+ batch_size: 4
23
+ lr: 0.0001
24
+ lr_wav2vec2: 0.00001
25
+
26
+ dataloader_options:
27
+ batch_size: !ref <batch_size>
28
+ shuffle: True
29
+ num_workers: 0
30
+ drop_last: False
31
+
32
+ encoder_dim: 768
33
+
34
+ # Number of emotions
35
+ out_n_neurons: 7 # (anger, disgust, fear, happy, neutral, sad, surprise)
hparams/train_with_wav2vec.yaml ADDED
@@ -0,0 +1,112 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ seed: 1993
2
+ __set_seed: !apply:torch.manual_seed [!ref <seed>]
3
+
4
+ # Dataset will be downloaded to the `data_original`
5
+ data_original: D:/voice-emo/dat/
6
+ output_folder: !ref results/train_with_wav2vec2/<seed>
7
+ save_folder: !ref <output_folder>/save
8
+ train_log: !ref <output_folder>/train_log.txt
9
+
10
+ # URL for the wav2vec2 model, you can change to benchmark different models
11
+ # Important: we use wav2vec2 base and not the fine-tuned one with ASR task
12
+ # This allows you to have ~4% improvement
13
+ wav2vec2_hub: facebook/wav2vec2-base
14
+ wav2vec2_folder: !ref <save_folder>/wav2vec2_checkpoint
15
+
16
+ # Path where data manifest files will be stored
17
+ train_annotation: !ref <output_folder>/train.json
18
+ valid_annotation: !ref <output_folder>/valid.json
19
+ test_annotation: !ref <output_folder>/test.json
20
+ split_ratio: [80, 10, 10]
21
+ skip_prep: False
22
+
23
+ # The train logger writes training statistics to a file, as well as stdout.
24
+ train_logger: !new:speechbrain.utils.train_logger.FileTrainLogger
25
+ save_file: !ref <train_log>
26
+
27
+ ckpt_interval_minutes: 15 # save checkpoint every N min
28
+
29
+ ####################### Training Parameters ####################################
30
+ number_of_epochs: 30
31
+ batch_size: 4
32
+ lr: 0.0001
33
+ lr_wav2vec2: 0.00001
34
+
35
+ # Freeze all wav2vec2
36
+ freeze_wav2vec2: False
37
+ # Set to true to freeze the CONV part of the wav2vec2 model
38
+ # We see an improvement of 2% with freezing CNNs
39
+ freeze_wav2vec2_conv: True
40
+
41
+ ####################### Model Parameters #######################################
42
+ encoder_dim: 768
43
+
44
+ # Number of emotions
45
+ out_n_neurons: 7 # (anger, disgust, fear, happy, neutral, sad, suprise )
46
+
47
+ dataloader_options:
48
+ batch_size: !ref <batch_size>
49
+ shuffle: True
50
+ num_workers: 2 # 2 on Linux but 0 works on Windows
51
+ drop_last: False
52
+
53
+ # Wav2vec2 encoder
54
+ wav2vec2: !new:speechbrain.lobes.models.huggingface_transformers.wav2vec2.Wav2Vec2
55
+ source: !ref <wav2vec2_hub>
56
+ output_norm: True
57
+ freeze: !ref <freeze_wav2vec2>
58
+ freeze_feature_extractor: !ref <freeze_wav2vec2_conv>
59
+ save_path: !ref <wav2vec2_folder>
60
+
61
+ avg_pool: !new:speechbrain.nnet.pooling.StatisticsPooling
62
+ return_std: False
63
+
64
+ output_mlp: !new:speechbrain.nnet.linear.Linear
65
+ input_size: !ref <encoder_dim>
66
+ n_neurons: !ref <out_n_neurons>
67
+ bias: False
68
+
69
+ epoch_counter: !new:speechbrain.utils.epoch_loop.EpochCounter
70
+ limit: !ref <number_of_epochs>
71
+
72
+ modules:
73
+ wav2vec2: !ref <wav2vec2>
74
+ output_mlp: !ref <output_mlp>
75
+
76
+ model: !new:torch.nn.ModuleList
77
+ - [!ref <output_mlp>]
78
+
79
+ log_softmax: !new:speechbrain.nnet.activations.Softmax
80
+ apply_log: True
81
+
82
+ compute_cost: !name:speechbrain.nnet.losses.nll_loss
83
+
84
+ error_stats: !name:speechbrain.utils.metric_stats.MetricStats
85
+ metric: !name:speechbrain.nnet.losses.classification_error
86
+ reduction: batch
87
+
88
+ opt_class: !name:torch.optim.Adam
89
+ lr: !ref <lr>
90
+
91
+ wav2vec2_opt_class: !name:torch.optim.Adam
92
+ lr: !ref <lr_wav2vec2>
93
+
94
+ lr_annealing: !new:speechbrain.nnet.schedulers.NewBobScheduler
95
+ initial_value: !ref <lr>
96
+ improvement_threshold: 0.0025
97
+ annealing_factor: 0.9
98
+ patient: 0
99
+
100
+ lr_annealing_wav2vec2: !new:speechbrain.nnet.schedulers.NewBobScheduler
101
+ initial_value: !ref <lr_wav2vec2>
102
+ improvement_threshold: 0.0025
103
+ annealing_factor: 0.9
104
+
105
+ checkpointer: !new:speechbrain.utils.checkpoints.Checkpointer
106
+ checkpoints_dir: !ref <save_folder>
107
+ recoverables:
108
+ model: !ref <model>
109
+ wav2vec2: !ref <wav2vec2>
110
+ lr_annealing_output: !ref <lr_annealing>
111
+ lr_annealing_wav2vec2: !ref <lr_annealing_wav2vec2>
112
+ counter: !ref <epoch_counter>
model.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6dd3e3ab14987cd5124407a58a263504ee5b7540727dadbdb04f6481f3775b1f
3
+ size 755087318
prepare.py ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import json
3
+ import random
4
+ import logging
5
+
6
+ logger = logging.getLogger(__name__)
7
+
8
+ def prepare_data(data_original, save_json_train, save_json_valid, save_json_test, split_ratio=[80, 10, 10], seed=12):
9
+ # Setting seeds for reproducible code.
10
+ random.seed(seed)
11
+
12
+ # Check if data preparation has already been done (skip if files exist)
13
+ if os.path.exists(save_json_train) and os.path.exists(save_json_valid) and os.path.exists(save_json_test):
14
+ logger.info("Preparation completed in previous run, skipping.")
15
+ return
16
+
17
+ # Collect audio files and labels
18
+ wav_list = []
19
+ labels = os.listdir(data_original)
20
+
21
+ for label in labels:
22
+ label_dir = os.path.join(data_original, label)
23
+ if os.path.isdir(label_dir):
24
+ for audio_file in os.listdir(label_dir):
25
+ if audio_file.endswith('.wav'):
26
+ wav_file = os.path.join(label_dir, audio_file)
27
+ if os.path.isfile(wav_file):
28
+ wav_list.append((wav_file, label))
29
+ else:
30
+ logger.warning(f"Skipping invalid audio file: {wav_file}")
31
+
32
+ # Shuffle and split the data
33
+ random.shuffle(wav_list)
34
+ n_total = len(wav_list)
35
+ n_train = n_total * split_ratio[0] // 100
36
+ n_valid = n_total * split_ratio[1] // 100
37
+
38
+ train_set = wav_list[:n_train]
39
+ valid_set = wav_list[n_train:n_train + n_valid]
40
+ test_set = wav_list[n_train + n_valid:]
41
+
42
+ # Create JSON files for train, valid, and test sets
43
+ create_json(train_set, save_json_train)
44
+ create_json(valid_set, save_json_valid)
45
+ create_json(test_set, save_json_test)
46
+
47
+ logger.info(f"Created {save_json_train}, {save_json_valid}, and {save_json_test}")
48
+
49
+ def create_json(data, json_file):
50
+ data_dict = {str(idx): {'wav': wav, 'label': label} for idx, (wav, label) in enumerate(data)}
51
+ with open(json_file, 'w') as f:
52
+ json.dump(data_dict, f)
results/train_with_wav2vec2/1993/test.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"0": {"wav": "D:/voice-emo/dat/surprise\\JK_su14.wav", "label": "surprise"}, "1": {"wav": "D:/voice-emo/dat/disgust\\1001_IWW_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "2": {"wav": "D:/voice-emo/dat/happy\\1001_IWL_HAP_XX.wav", "label": "happy"}, "3": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_HI_noise_augmented.wav", "label": "disgust"}, "4": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_LO.wav", "label": "sad"}, "5": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_LO_noise_augmented.wav", "label": "sad"}, "6": {"wav": "D:/voice-emo/dat/fear\\1001_IOM_FEA_XX_pitch_augmented.wav", "label": "fear"}, "7": {"wav": "D:/voice-emo/dat/neutral\\1001_IEO_NEU_XX.wav", "label": "neutral"}, "8": {"wav": "D:/voice-emo/dat/disgust\\1001_TAI_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "9": {"wav": "D:/voice-emo/dat/fear\\1001_WSI_FEA_XX_noise_augmented.wav", "label": "fear"}, "10": {"wav": "D:/voice-emo/dat/angry\\1001_IOM_ANG_XX.wav", "label": "angry"}, "11": {"wav": "D:/voice-emo/dat/angry\\1001_WSI_ANG_XX_stretch_augmented.wav", "label": "angry"}, "12": {"wav": "D:/voice-emo/dat/disgust\\1001_IOM_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "13": {"wav": "D:/voice-emo/dat/angry\\1001_MTI_ANG_XX.wav", "label": "angry"}, "14": {"wav": "D:/voice-emo/dat/surprise\\DC_su01_noise_augmented.wav", "label": "surprise"}, "15": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_HI_stretch_augmented.wav", "label": "fear"}, "16": {"wav": "D:/voice-emo/dat/neutral\\1001_TIE_NEU_XX_noise_augmented.wav", "label": "neutral"}, "17": {"wav": "D:/voice-emo/dat/disgust\\1001_TAI_DIS_XX_noise_augmented.wav", "label": "disgust"}, "18": {"wav": "D:/voice-emo/dat/angry\\1001_DFA_ANG_XX_stretch_augmented.wav", "label": "angry"}, "19": {"wav": "D:/voice-emo/dat/neutral\\1001_DFA_NEU_XX.wav", "label": "neutral"}, "20": {"wav": "D:/voice-emo/dat/surprise\\DC_su03.wav", "label": "surprise"}, "21": {"wav": "D:/voice-emo/dat/fear\\1001_TIE_FEA_XX_stretch_augmented.wav", "label": "fear"}, "22": {"wav": "D:/voice-emo/dat/fear\\1001_IWL_FEA_XX_noise_augmented.wav", "label": "fear"}, "23": {"wav": "D:/voice-emo/dat/fear\\1001_IWL_FEA_XX_stretch_augmented.wav", "label": "fear"}, "24": {"wav": "D:/voice-emo/dat/happy\\1001_MTI_HAP_XX_pitch_augmented.wav", "label": "happy"}, "25": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_MD_noise_augmented.wav", "label": "happy"}, "26": {"wav": "D:/voice-emo/dat/sad\\1001_ITH_SAD_XX_stretch_augmented.wav", "label": "sad"}, "27": {"wav": "D:/voice-emo/dat/happy\\1001_MTI_HAP_XX.wav", "label": "happy"}, "28": {"wav": "D:/voice-emo/dat/disgust\\1001_ITH_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "29": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_LO_stretch_augmented.wav", "label": "happy"}, "30": {"wav": "D:/voice-emo/dat/happy\\1001_DFA_HAP_XX_pitch_augmented.wav", "label": "happy"}, "31": {"wav": "D:/voice-emo/dat/disgust\\1001_IWL_DIS_XX_noise_augmented.wav", "label": "disgust"}, "32": {"wav": "D:/voice-emo/dat/disgust\\1001_DFA_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "33": {"wav": "D:/voice-emo/dat/angry\\1001_IOM_ANG_XX_pitch_augmented.wav", "label": "angry"}, "34": {"wav": "D:/voice-emo/dat/neutral\\1001_IEO_NEU_XX_noise_augmented.wav", "label": "neutral"}, "35": {"wav": "D:/voice-emo/dat/fear\\1001_ITH_FEA_XX_pitch_augmented.wav", "label": "fear"}, "36": {"wav": "D:/voice-emo/dat/sad\\1001_TIE_SAD_XX_noise_augmented.wav", "label": "sad"}, "37": {"wav": "D:/voice-emo/dat/neutral\\1001_IWL_NEU_XX_pitch_augmented.wav", "label": "neutral"}}
results/train_with_wav2vec2/1993/train.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"0": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_HI_noise_augmented.wav", "label": "angry"}, "1": {"wav": "D:/voice-emo/dat/angry\\1001_DFA_ANG_XX_pitch_augmented.wav", "label": "angry"}, "2": {"wav": "D:/voice-emo/dat/angry\\1001_ITH_ANG_XX.wav", "label": "angry"}, "3": {"wav": "D:/voice-emo/dat/fear\\1001_IWW_FEA_XX_noise_augmented.wav", "label": "fear"}, "4": {"wav": "D:/voice-emo/dat/neutral\\1001_MTI_NEU_XX.wav", "label": "neutral"}, "5": {"wav": "D:/voice-emo/dat/surprise\\DC_su08_noise_augmented.wav", "label": "surprise"}, "6": {"wav": "D:/voice-emo/dat/surprise\\DC_su04.wav", "label": "surprise"}, "7": {"wav": "D:/voice-emo/dat/surprise\\DC_su01_pitch_augmented.wav", "label": "surprise"}, "8": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_MD_pitch_augmented.wav", "label": "angry"}, "9": {"wav": "D:/voice-emo/dat/fear\\1001_DFA_FEA_XX_stretch_augmented.wav", "label": "fear"}, "10": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_HI_stretch_augmented.wav", "label": "disgust"}, "11": {"wav": "D:/voice-emo/dat/surprise\\DC_su07_noise_augmented.wav", "label": "surprise"}, "12": {"wav": "D:/voice-emo/dat/neutral\\1001_ITH_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "13": {"wav": "D:/voice-emo/dat/fear\\1001_TIE_FEA_XX.wav", "label": "fear"}, "14": {"wav": "D:/voice-emo/dat/happy\\1001_TAI_HAP_XX_noise_augmented.wav", "label": "happy"}, "15": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_LO_pitch_augmented.wav", "label": "happy"}, "16": {"wav": "D:/voice-emo/dat/neutral\\1001_ITH_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "17": {"wav": "D:/voice-emo/dat/sad\\1001_ITS_SAD_XX_noise_augmented.wav", "label": "sad"}, "18": {"wav": "D:/voice-emo/dat/neutral\\1001_ITS_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "19": {"wav": "D:/voice-emo/dat/sad\\1001_TAI_SAD_XX_pitch_augmented.wav", "label": "sad"}, "20": {"wav": "D:/voice-emo/dat/sad\\1001_ITH_SAD_XX_pitch_augmented.wav", "label": "sad"}, "21": {"wav": "D:/voice-emo/dat/angry\\1001_ITS_ANG_XX.wav", "label": "angry"}, "22": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_HI.wav", "label": "happy"}, "23": {"wav": "D:/voice-emo/dat/sad\\1001_IOM_SAD_XX_pitch_augmented.wav", "label": "sad"}, "24": {"wav": "D:/voice-emo/dat/sad\\1001_TSI_SAD_XX_stretch_augmented.wav", "label": "sad"}, "25": {"wav": "D:/voice-emo/dat/surprise\\DC_su08.wav", "label": "surprise"}, "26": {"wav": "D:/voice-emo/dat/sad\\1001_TSI_SAD_XX.wav", "label": "sad"}, "27": {"wav": "D:/voice-emo/dat/sad\\1001_TSI_SAD_XX_noise_augmented.wav", "label": "sad"}, "28": {"wav": "D:/voice-emo/dat/neutral\\1001_TSI_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "29": {"wav": "D:/voice-emo/dat/angry\\1001_TSI_ANG_XX_pitch_augmented.wav", "label": "angry"}, "30": {"wav": "D:/voice-emo/dat/fear\\1001_ITH_FEA_XX_noise_augmented.wav", "label": "fear"}, "31": {"wav": "D:/voice-emo/dat/neutral\\1001_ITS_NEU_XX_noise_augmented.wav", "label": "neutral"}, "32": {"wav": "D:/voice-emo/dat/neutral\\1001_TAI_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "33": {"wav": "D:/voice-emo/dat/angry\\1001_TSI_ANG_XX_noise_augmented.wav", "label": "angry"}, "34": {"wav": "D:/voice-emo/dat/disgust\\1001_ITS_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "35": {"wav": "D:/voice-emo/dat/surprise\\DC_su04_noise_augmented.wav", "label": "surprise"}, "36": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_MD_noise_augmented.wav", "label": "sad"}, "37": {"wav": "D:/voice-emo/dat/surprise\\DC_su08_pitch_augmented.wav", "label": "surprise"}, "38": {"wav": "D:/voice-emo/dat/sad\\1001_IWL_SAD_XX_pitch_augmented.wav", "label": "sad"}, "39": {"wav": "D:/voice-emo/dat/fear\\1001_DFA_FEA_XX.wav", "label": "fear"}, "40": {"wav": "D:/voice-emo/dat/happy\\1001_ITH_HAP_XX.wav", "label": "happy"}, "41": {"wav": "D:/voice-emo/dat/disgust\\1001_MTI_DIS_XX.wav", "label": "disgust"}, "42": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_LO_noise_augmented.wav", "label": "happy"}, "43": {"wav": "D:/voice-emo/dat/surprise\\DC_su03_stretch_augmented.wav", "label": "surprise"}, "44": {"wav": "D:/voice-emo/dat/surprise\\DC_su04_pitch_augmented.wav", "label": "surprise"}, "45": {"wav": "D:/voice-emo/dat/surprise\\DC_su07_pitch_augmented.wav", "label": "surprise"}, "46": {"wav": "D:/voice-emo/dat/happy\\1001_IWW_HAP_XX_stretch_augmented.wav", "label": "happy"}, "47": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_MD.wav", "label": "fear"}, "48": {"wav": "D:/voice-emo/dat/angry\\1001_TSI_ANG_XX_stretch_augmented.wav", "label": "angry"}, "49": {"wav": "D:/voice-emo/dat/neutral\\1001_IOM_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "50": {"wav": "D:/voice-emo/dat/fear\\1001_IOM_FEA_XX_noise_augmented.wav", "label": "fear"}, "51": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_LO_noise_augmented.wav", "label": "disgust"}, "52": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_MD_noise_augmented.wav", "label": "disgust"}, "53": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_MD_stretch_augmented.wav", "label": "disgust"}, "54": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_LO_pitch_augmented.wav", "label": "fear"}, "55": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_LO_pitch_augmented.wav", "label": "disgust"}, "56": {"wav": "D:/voice-emo/dat/disgust\\1001_IWW_DIS_XX.wav", "label": "disgust"}, "57": {"wav": "D:/voice-emo/dat/angry\\1001_IWL_ANG_XX_noise_augmented.wav", "label": "angry"}, "58": {"wav": "D:/voice-emo/dat/happy\\1001_TAI_HAP_XX_stretch_augmented.wav", "label": "happy"}, "59": {"wav": "D:/voice-emo/dat/neutral\\1001_ITH_NEU_XX_noise_augmented.wav", "label": "neutral"}, "60": {"wav": "D:/voice-emo/dat/happy\\1001_IOM_HAP_XX_stretch_augmented.wav", "label": "happy"}, "61": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_HI_pitch_augmented.wav", "label": "sad"}, "62": {"wav": "D:/voice-emo/dat/sad\\1001_TIE_SAD_XX.wav", "label": "sad"}, "63": {"wav": "D:/voice-emo/dat/angry\\1001_MTI_ANG_XX_stretch_augmented.wav", "label": "angry"}, "64": {"wav": "D:/voice-emo/dat/disgust\\1001_ITH_DIS_XX_noise_augmented.wav", "label": "disgust"}, "65": {"wav": "D:/voice-emo/dat/neutral\\1001_TAI_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "66": {"wav": "D:/voice-emo/dat/fear\\1001_ITH_FEA_XX.wav", "label": "fear"}, "67": {"wav": "D:/voice-emo/dat/surprise\\DC_su09.wav", "label": "surprise"}, "68": {"wav": "D:/voice-emo/dat/sad\\1001_DFA_SAD_XX_pitch_augmented.wav", "label": "sad"}, "69": {"wav": "D:/voice-emo/dat/surprise\\DC_su05_stretch_augmented.wav", "label": "surprise"}, "70": {"wav": "D:/voice-emo/dat/neutral\\1001_IWL_NEU_XX_noise_augmented.wav", "label": "neutral"}, "71": {"wav": "D:/voice-emo/dat/disgust\\1001_IOM_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "72": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_LO_noise_augmented.wav", "label": "angry"}, "73": {"wav": "D:/voice-emo/dat/sad\\1001_IWW_SAD_XX_pitch_augmented.wav", "label": "sad"}, "74": {"wav": "D:/voice-emo/dat/angry\\1001_ITS_ANG_XX_stretch_augmented.wav", "label": "angry"}, "75": {"wav": "D:/voice-emo/dat/sad\\1001_WSI_SAD_XX_stretch_augmented.wav", "label": "sad"}, "76": {"wav": "D:/voice-emo/dat/neutral\\1001_IEO_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "77": {"wav": "D:/voice-emo/dat/neutral\\1001_WSI_NEU_XX.wav", "label": "neutral"}, "78": {"wav": "D:/voice-emo/dat/disgust\\1001_MTI_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "79": {"wav": "D:/voice-emo/dat/disgust\\1001_WSI_DIS_XX_noise_augmented.wav", "label": "disgust"}, "80": {"wav": "D:/voice-emo/dat/neutral\\1001_IOM_NEU_XX.wav", "label": "neutral"}, "81": {"wav": "D:/voice-emo/dat/surprise\\DC_su04_stretch_augmented.wav", "label": "surprise"}, "82": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_LO_pitch_augmented.wav", "label": "sad"}, "83": {"wav": "D:/voice-emo/dat/fear\\1001_WSI_FEA_XX.wav", "label": "fear"}, "84": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_LO_stretch_augmented.wav", "label": "disgust"}, "85": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_HI_pitch_augmented.wav", "label": "fear"}, "86": {"wav": "D:/voice-emo/dat/neutral\\1001_IWL_NEU_XX.wav", "label": "neutral"}, "87": {"wav": "D:/voice-emo/dat/neutral\\1001_IWW_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "88": {"wav": "D:/voice-emo/dat/angry\\1001_WSI_ANG_XX_pitch_augmented.wav", "label": "angry"}, "89": {"wav": "D:/voice-emo/dat/angry\\1001_ITH_ANG_XX_pitch_augmented.wav", "label": "angry"}, "90": {"wav": "D:/voice-emo/dat/happy\\1001_TIE_HAP_XX_pitch_augmented.wav", "label": "happy"}, "91": {"wav": "D:/voice-emo/dat/neutral\\1001_TAI_NEU_XX.wav", "label": "neutral"}, "92": {"wav": "D:/voice-emo/dat/disgust\\1001_IOM_DIS_XX_noise_augmented.wav", "label": "disgust"}, "93": {"wav": "D:/voice-emo/dat/angry\\1001_IWL_ANG_XX_stretch_augmented.wav", "label": "angry"}, "94": {"wav": "D:/voice-emo/dat/fear\\1001_ITS_FEA_XX_stretch_augmented.wav", "label": "fear"}, "95": {"wav": "D:/voice-emo/dat/surprise\\DC_su06_pitch_augmented.wav", "label": "surprise"}, "96": {"wav": "D:/voice-emo/dat/sad\\1001_TAI_SAD_XX.wav", "label": "sad"}, "97": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_HI_noise_augmented.wav", "label": "sad"}, "98": {"wav": "D:/voice-emo/dat/surprise\\DC_su06_stretch_augmented.wav", "label": "surprise"}, "99": {"wav": "D:/voice-emo/dat/angry\\1001_TSI_ANG_XX.wav", "label": "angry"}, "100": {"wav": "D:/voice-emo/dat/neutral\\1001_TIE_NEU_XX.wav", "label": "neutral"}, "101": {"wav": "D:/voice-emo/dat/sad\\1001_MTI_SAD_XX_pitch_augmented.wav", "label": "sad"}, "102": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_MD_stretch_augmented.wav", "label": "angry"}, "103": {"wav": "D:/voice-emo/dat/surprise\\DC_su09_pitch_augmented.wav", "label": "surprise"}, "104": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_LO_stretch_augmented.wav", "label": "fear"}, "105": {"wav": "D:/voice-emo/dat/fear\\1001_IWL_FEA_XX_pitch_augmented.wav", "label": "fear"}, "106": {"wav": "D:/voice-emo/dat/angry\\1001_ITH_ANG_XX_noise_augmented.wav", "label": "angry"}, "107": {"wav": "D:/voice-emo/dat/sad\\1001_IWL_SAD_XX.wav", "label": "sad"}, "108": {"wav": "D:/voice-emo/dat/disgust\\1001_TIE_DIS_XX.wav", "label": "disgust"}, "109": {"wav": "D:/voice-emo/dat/fear\\1001_TAI_FEA_XX_stretch_augmented.wav", "label": "fear"}, "110": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_MD_noise_augmented.wav", "label": "angry"}, "111": {"wav": "D:/voice-emo/dat/neutral\\1001_MTI_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "112": {"wav": "D:/voice-emo/dat/disgust\\1001_TAI_DIS_XX.wav", "label": "disgust"}, "113": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_HI_noise_augmented.wav", "label": "fear"}, "114": {"wav": "D:/voice-emo/dat/disgust\\1001_IOM_DIS_XX.wav", "label": "disgust"}, "115": {"wav": "D:/voice-emo/dat/sad\\1001_TIE_SAD_XX_pitch_augmented.wav", "label": "sad"}, "116": {"wav": "D:/voice-emo/dat/disgust\\1001_TSI_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "117": {"wav": "D:/voice-emo/dat/disgust\\1001_TSI_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "118": {"wav": "D:/voice-emo/dat/disgust\\1001_IWW_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "119": {"wav": "D:/voice-emo/dat/angry\\1001_MTI_ANG_XX_pitch_augmented.wav", "label": "angry"}, "120": {"wav": "D:/voice-emo/dat/sad\\1001_TAI_SAD_XX_noise_augmented.wav", "label": "sad"}, "121": {"wav": "D:/voice-emo/dat/disgust\\1001_TIE_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "122": {"wav": "D:/voice-emo/dat/surprise\\DC_su02_stretch_augmented.wav", "label": "surprise"}, "123": {"wav": "D:/voice-emo/dat/surprise\\DC_su10_noise_augmented.wav", "label": "surprise"}, "124": {"wav": "D:/voice-emo/dat/disgust\\1001_WSI_DIS_XX.wav", "label": "disgust"}, "125": {"wav": "D:/voice-emo/dat/happy\\1001_WSI_HAP_XX_noise_augmented.wav", "label": "happy"}, "126": {"wav": "D:/voice-emo/dat/angry\\1001_DFA_ANG_XX_noise_augmented.wav", "label": "angry"}, "127": {"wav": "D:/voice-emo/dat/fear\\1001_ITS_FEA_XX_noise_augmented.wav", "label": "fear"}, "128": {"wav": "D:/voice-emo/dat/happy\\1001_TSI_HAP_XX_pitch_augmented.wav", "label": "happy"}, "129": {"wav": "D:/voice-emo/dat/happy\\1001_ITS_HAP_XX_stretch_augmented.wav", "label": "happy"}, "130": {"wav": "D:/voice-emo/dat/sad\\1001_WSI_SAD_XX.wav", "label": "sad"}, "131": {"wav": "D:/voice-emo/dat/fear\\1001_TAI_FEA_XX_noise_augmented.wav", "label": "fear"}, "132": {"wav": "D:/voice-emo/dat/angry\\1001_DFA_ANG_XX.wav", "label": "angry"}, "133": {"wav": "D:/voice-emo/dat/sad\\1001_WSI_SAD_XX_pitch_augmented.wav", "label": "sad"}, "134": {"wav": "D:/voice-emo/dat/angry\\1001_MTI_ANG_XX_noise_augmented.wav", "label": "angry"}, "135": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_MD.wav", "label": "disgust"}, "136": {"wav": "D:/voice-emo/dat/sad\\1001_MTI_SAD_XX_noise_augmented.wav", "label": "sad"}, "137": {"wav": "D:/voice-emo/dat/neutral\\1001_DFA_NEU_XX_noise_augmented.wav", "label": "neutral"}, "138": {"wav": "D:/voice-emo/dat/fear\\1001_MTI_FEA_XX.wav", "label": "fear"}, "139": {"wav": "D:/voice-emo/dat/sad\\1001_TSI_SAD_XX_pitch_augmented.wav", "label": "sad"}, "140": {"wav": "D:/voice-emo/dat/disgust\\1001_ITS_DIS_XX.wav", "label": "disgust"}, "141": {"wav": "D:/voice-emo/dat/neutral\\1001_WSI_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "142": {"wav": "D:/voice-emo/dat/fear\\1001_TAI_FEA_XX_pitch_augmented.wav", "label": "fear"}, "143": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_MD_stretch_augmented.wav", "label": "sad"}, "144": {"wav": "D:/voice-emo/dat/angry\\1001_ITS_ANG_XX_pitch_augmented.wav", "label": "angry"}, "145": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_HI_stretch_augmented.wav", "label": "sad"}, "146": {"wav": "D:/voice-emo/dat/happy\\1001_ITS_HAP_XX_noise_augmented.wav", "label": "happy"}, "147": {"wav": "D:/voice-emo/dat/angry\\1001_TIE_ANG_XX_stretch_augmented.wav", "label": "angry"}, "148": {"wav": "D:/voice-emo/dat/happy\\1001_TIE_HAP_XX.wav", "label": "happy"}, "149": {"wav": "D:/voice-emo/dat/fear\\1001_WSI_FEA_XX_stretch_augmented.wav", "label": "fear"}, "150": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_LO.wav", "label": "disgust"}, "151": {"wav": "D:/voice-emo/dat/sad\\1001_IOM_SAD_XX.wav", "label": "sad"}, "152": {"wav": "D:/voice-emo/dat/sad\\1001_MTI_SAD_XX_stretch_augmented.wav", "label": "sad"}, "153": {"wav": "D:/voice-emo/dat/happy\\1001_IOM_HAP_XX_pitch_augmented.wav", "label": "happy"}, "154": {"wav": "D:/voice-emo/dat/happy\\1001_IOM_HAP_XX_noise_augmented.wav", "label": "happy"}, "155": {"wav": "D:/voice-emo/dat/sad\\1001_MTI_SAD_XX.wav", "label": "sad"}, "156": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_MD_noise_augmented.wav", "label": "fear"}, "157": {"wav": "D:/voice-emo/dat/sad\\1001_ITS_SAD_XX_stretch_augmented.wav", "label": "sad"}, "158": {"wav": "D:/voice-emo/dat/sad\\1001_IWL_SAD_XX_noise_augmented.wav", "label": "sad"}, "159": {"wav": "D:/voice-emo/dat/neutral\\1001_IWW_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "160": {"wav": "D:/voice-emo/dat/angry\\1001_ITH_ANG_XX_stretch_augmented.wav", "label": "angry"}, "161": {"wav": "D:/voice-emo/dat/happy\\1001_MTI_HAP_XX_noise_augmented.wav", "label": "happy"}, "162": {"wav": "D:/voice-emo/dat/angry\\1001_WSI_ANG_XX.wav", "label": "angry"}, "163": {"wav": "D:/voice-emo/dat/neutral\\1001_TIE_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "164": {"wav": "D:/voice-emo/dat/sad\\1001_WSI_SAD_XX_noise_augmented.wav", "label": "sad"}, "165": {"wav": "D:/voice-emo/dat/angry\\1001_IWW_ANG_XX_pitch_augmented.wav", "label": "angry"}, "166": {"wav": "D:/voice-emo/dat/happy\\1001_IWW_HAP_XX.wav", "label": "happy"}, "167": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_HI_pitch_augmented.wav", "label": "angry"}, "168": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_MD_stretch_augmented.wav", "label": "happy"}, "169": {"wav": "D:/voice-emo/dat/sad\\1001_ITH_SAD_XX.wav", "label": "sad"}, "170": {"wav": "D:/voice-emo/dat/happy\\1001_TAI_HAP_XX.wav", "label": "happy"}, "171": {"wav": "D:/voice-emo/dat/fear\\1001_IWW_FEA_XX_pitch_augmented.wav", "label": "fear"}, "172": {"wav": "D:/voice-emo/dat/sad\\1001_ITS_SAD_XX.wav", "label": "sad"}, "173": {"wav": "D:/voice-emo/dat/angry\\1001_TIE_ANG_XX.wav", "label": "angry"}, "174": {"wav": "D:/voice-emo/dat/disgust\\1001_MTI_DIS_XX_noise_augmented.wav", "label": "disgust"}, "175": {"wav": "D:/voice-emo/dat/surprise\\DC_su06.wav", "label": "surprise"}, "176": {"wav": "D:/voice-emo/dat/angry\\1001_ITS_ANG_XX_noise_augmented.wav", "label": "angry"}, "177": {"wav": "D:/voice-emo/dat/neutral\\1001_IWL_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "178": {"wav": "D:/voice-emo/dat/neutral\\1001_ITS_NEU_XX.wav", "label": "neutral"}, "179": {"wav": "D:/voice-emo/dat/disgust\\1001_IWW_DIS_XX_noise_augmented.wav", "label": "disgust"}, "180": {"wav": "D:/voice-emo/dat/neutral\\1001_MTI_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "181": {"wav": "D:/voice-emo/dat/sad\\1001_IWW_SAD_XX.wav", "label": "sad"}, "182": {"wav": "D:/voice-emo/dat/fear\\1001_TSI_FEA_XX_pitch_augmented.wav", "label": "fear"}, "183": {"wav": "D:/voice-emo/dat/surprise\\DC_su08_stretch_augmented.wav", "label": "surprise"}, "184": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_HI.wav", "label": "fear"}, "185": {"wav": "D:/voice-emo/dat/happy\\1001_ITS_HAP_XX_pitch_augmented.wav", "label": "happy"}, "186": {"wav": "D:/voice-emo/dat/surprise\\DC_su05_pitch_augmented.wav", "label": "surprise"}, "187": {"wav": "D:/voice-emo/dat/fear\\1001_IWW_FEA_XX.wav", "label": "fear"}, "188": {"wav": "D:/voice-emo/dat/disgust\\1001_TSI_DIS_XX.wav", "label": "disgust"}, "189": {"wav": "D:/voice-emo/dat/neutral\\1001_TSI_NEU_XX.wav", "label": "neutral"}, "190": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_LO_noise_augmented.wav", "label": "fear"}, "191": {"wav": "D:/voice-emo/dat/happy\\1001_TIE_HAP_XX_noise_augmented.wav", "label": "happy"}, "192": {"wav": "D:/voice-emo/dat/happy\\1001_DFA_HAP_XX.wav", "label": "happy"}, "193": {"wav": "D:/voice-emo/dat/sad\\1001_DFA_SAD_XX_noise_augmented.wav", "label": "sad"}, "194": {"wav": "D:/voice-emo/dat/fear\\1001_TAI_FEA_XX.wav", "label": "fear"}, "195": {"wav": "D:/voice-emo/dat/angry\\1001_IOM_ANG_XX_stretch_augmented.wav", "label": "angry"}, "196": {"wav": "D:/voice-emo/dat/disgust\\1001_ITS_DIS_XX_noise_augmented.wav", "label": "disgust"}, "197": {"wav": "D:/voice-emo/dat/disgust\\1001_DFA_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "198": {"wav": "D:/voice-emo/dat/surprise\\DC_su09_noise_augmented.wav", "label": "surprise"}, "199": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_HI.wav", "label": "sad"}, "200": {"wav": "D:/voice-emo/dat/disgust\\1001_WSI_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "201": {"wav": "D:/voice-emo/dat/fear\\1001_WSI_FEA_XX_pitch_augmented.wav", "label": "fear"}, "202": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_HI_pitch_augmented.wav", "label": "disgust"}, "203": {"wav": "D:/voice-emo/dat/surprise\\DC_su02_pitch_augmented.wav", "label": "surprise"}, "204": {"wav": "D:/voice-emo/dat/fear\\1001_ITS_FEA_XX_pitch_augmented.wav", "label": "fear"}, "205": {"wav": "D:/voice-emo/dat/fear\\1001_TSI_FEA_XX_stretch_augmented.wav", "label": "fear"}, "206": {"wav": "D:/voice-emo/dat/happy\\1001_ITS_HAP_XX.wav", "label": "happy"}, "207": {"wav": "D:/voice-emo/dat/disgust\\1001_IWL_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "208": {"wav": "D:/voice-emo/dat/fear\\1001_IWL_FEA_XX.wav", "label": "fear"}, "209": {"wav": "D:/voice-emo/dat/happy\\1001_WSI_HAP_XX.wav", "label": "happy"}, "210": {"wav": "D:/voice-emo/dat/angry\\1001_TAI_ANG_XX_pitch_augmented.wav", "label": "angry"}, "211": {"wav": "D:/voice-emo/dat/disgust\\1001_TSI_DIS_XX_noise_augmented.wav", "label": "disgust"}, "212": {"wav": "D:/voice-emo/dat/happy\\1001_IWL_HAP_XX_noise_augmented.wav", "label": "happy"}, "213": {"wav": "D:/voice-emo/dat/happy\\1001_TIE_HAP_XX_stretch_augmented.wav", "label": "happy"}, "214": {"wav": "D:/voice-emo/dat/surprise\\DC_su06_noise_augmented.wav", "label": "surprise"}, "215": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_MD_pitch_augmented.wav", "label": "fear"}, "216": {"wav": "D:/voice-emo/dat/angry\\1001_IWW_ANG_XX_stretch_augmented.wav", "label": "angry"}, "217": {"wav": "D:/voice-emo/dat/disgust\\1001_DFA_DIS_XX.wav", "label": "disgust"}, "218": {"wav": "D:/voice-emo/dat/fear\\1001_TSI_FEA_XX_noise_augmented.wav", "label": "fear"}, "219": {"wav": "D:/voice-emo/dat/disgust\\1001_TAI_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "220": {"wav": "D:/voice-emo/dat/angry\\1001_IWW_ANG_XX_noise_augmented.wav", "label": "angry"}, "221": {"wav": "D:/voice-emo/dat/sad\\1001_IOM_SAD_XX_stretch_augmented.wav", "label": "sad"}, "222": {"wav": "D:/voice-emo/dat/fear\\1001_DFA_FEA_XX_noise_augmented.wav", "label": "fear"}, "223": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_LO.wav", "label": "fear"}, "224": {"wav": "D:/voice-emo/dat/surprise\\DC_su07.wav", "label": "surprise"}, "225": {"wav": "D:/voice-emo/dat/sad\\1001_DFA_SAD_XX.wav", "label": "sad"}, "226": {"wav": "D:/voice-emo/dat/fear\\1001_MTI_FEA_XX_pitch_augmented.wav", "label": "fear"}, "227": {"wav": "D:/voice-emo/dat/neutral\\1001_ITH_NEU_XX.wav", "label": "neutral"}, "228": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_LO_pitch_augmented.wav", "label": "angry"}, "229": {"wav": "D:/voice-emo/dat/surprise\\DC_su10.wav", "label": "surprise"}, "230": {"wav": "D:/voice-emo/dat/disgust\\1001_DFA_DIS_XX_noise_augmented.wav", "label": "disgust"}, "231": {"wav": "D:/voice-emo/dat/disgust\\1001_IWL_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "232": {"wav": "D:/voice-emo/dat/disgust\\1001_ITH_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "233": {"wav": "D:/voice-emo/dat/fear\\1001_IWW_FEA_XX_stretch_augmented.wav", "label": "fear"}, "234": {"wav": "D:/voice-emo/dat/neutral\\1001_IWW_NEU_XX_noise_augmented.wav", "label": "neutral"}, "235": {"wav": "D:/voice-emo/dat/sad\\1001_ITS_SAD_XX_pitch_augmented.wav", "label": "sad"}, "236": {"wav": "D:/voice-emo/dat/angry\\1001_IWW_ANG_XX.wav", "label": "angry"}, "237": {"wav": "D:/voice-emo/dat/surprise\\DC_su01_stretch_augmented.wav", "label": "surprise"}, "238": {"wav": "D:/voice-emo/dat/fear\\1001_TSI_FEA_XX.wav", "label": "fear"}, "239": {"wav": "D:/voice-emo/dat/neutral\\1001_IEO_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "240": {"wav": "D:/voice-emo/dat/sad\\1001_IWW_SAD_XX_stretch_augmented.wav", "label": "sad"}, "241": {"wav": "D:/voice-emo/dat/surprise\\DC_su01.wav", "label": "surprise"}, "242": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_LO_stretch_augmented.wav", "label": "angry"}, "243": {"wav": "D:/voice-emo/dat/disgust\\1001_TIE_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "244": {"wav": "D:/voice-emo/dat/sad\\1001_IOM_SAD_XX_noise_augmented.wav", "label": "sad"}, "245": {"wav": "D:/voice-emo/dat/fear\\1001_IOM_FEA_XX.wav", "label": "fear"}, "246": {"wav": "D:/voice-emo/dat/sad\\1001_TAI_SAD_XX_stretch_augmented.wav", "label": "sad"}, "247": {"wav": "D:/voice-emo/dat/disgust\\1001_TIE_DIS_XX_noise_augmented.wav", "label": "disgust"}, "248": {"wav": "D:/voice-emo/dat/disgust\\1001_WSI_DIS_XX_stretch_augmented.wav", "label": "disgust"}, "249": {"wav": "D:/voice-emo/dat/sad\\1001_IWL_SAD_XX_stretch_augmented.wav", "label": "sad"}, "250": {"wav": "D:/voice-emo/dat/happy\\1001_TSI_HAP_XX.wav", "label": "happy"}, "251": {"wav": "D:/voice-emo/dat/fear\\1001_ITH_FEA_XX_stretch_augmented.wav", "label": "fear"}, "252": {"wav": "D:/voice-emo/dat/fear\\1001_TIE_FEA_XX_pitch_augmented.wav", "label": "fear"}, "253": {"wav": "D:/voice-emo/dat/angry\\1001_WSI_ANG_XX_noise_augmented.wav", "label": "angry"}, "254": {"wav": "D:/voice-emo/dat/angry\\1001_TAI_ANG_XX_noise_augmented.wav", "label": "angry"}, "255": {"wav": "D:/voice-emo/dat/happy\\1001_WSI_HAP_XX_stretch_augmented.wav", "label": "happy"}, "256": {"wav": "D:/voice-emo/dat/neutral\\1001_TAI_NEU_XX_noise_augmented.wav", "label": "neutral"}, "257": {"wav": "D:/voice-emo/dat/surprise\\DC_su03_noise_augmented.wav", "label": "surprise"}, "258": {"wav": "D:/voice-emo/dat/happy\\1001_MTI_HAP_XX_stretch_augmented.wav", "label": "happy"}, "259": {"wav": "D:/voice-emo/dat/neutral\\1001_IOM_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "260": {"wav": "D:/voice-emo/dat/happy\\1001_WSI_HAP_XX_pitch_augmented.wav", "label": "happy"}, "261": {"wav": "D:/voice-emo/dat/happy\\1001_ITH_HAP_XX_stretch_augmented.wav", "label": "happy"}, "262": {"wav": "D:/voice-emo/dat/sad\\1001_ITH_SAD_XX_noise_augmented.wav", "label": "sad"}, "263": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_LO_stretch_augmented.wav", "label": "sad"}, "264": {"wav": "D:/voice-emo/dat/angry\\1001_TIE_ANG_XX_pitch_augmented.wav", "label": "angry"}, "265": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_HI_stretch_augmented.wav", "label": "happy"}, "266": {"wav": "D:/voice-emo/dat/neutral\\1001_TSI_NEU_XX_noise_augmented.wav", "label": "neutral"}, "267": {"wav": "D:/voice-emo/dat/happy\\1001_IWW_HAP_XX_noise_augmented.wav", "label": "happy"}, "268": {"wav": "D:/voice-emo/dat/angry\\1001_IWL_ANG_XX.wav", "label": "angry"}, "269": {"wav": "D:/voice-emo/dat/surprise\\DC_su09_stretch_augmented.wav", "label": "surprise"}, "270": {"wav": "D:/voice-emo/dat/surprise\\DC_su10_pitch_augmented.wav", "label": "surprise"}, "271": {"wav": "D:/voice-emo/dat/neutral\\1001_WSI_NEU_XX_noise_augmented.wav", "label": "neutral"}, "272": {"wav": "D:/voice-emo/dat/surprise\\DC_su05_noise_augmented.wav", "label": "surprise"}, "273": {"wav": "D:/voice-emo/dat/angry\\1001_TAI_ANG_XX_stretch_augmented.wav", "label": "angry"}, "274": {"wav": "D:/voice-emo/dat/angry\\1001_TAI_ANG_XX.wav", "label": "angry"}, "275": {"wav": "D:/voice-emo/dat/happy\\1001_TAI_HAP_XX_pitch_augmented.wav", "label": "happy"}, "276": {"wav": "D:/voice-emo/dat/fear\\1001_ITS_FEA_XX.wav", "label": "fear"}, "277": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_LO.wav", "label": "happy"}, "278": {"wav": "D:/voice-emo/dat/surprise\\DC_su02_noise_augmented.wav", "label": "surprise"}, "279": {"wav": "D:/voice-emo/dat/neutral\\1001_IWW_NEU_XX.wav", "label": "neutral"}, "280": {"wav": "D:/voice-emo/dat/neutral\\1001_DFA_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "281": {"wav": "D:/voice-emo/dat/happy\\1001_DFA_HAP_XX_stretch_augmented.wav", "label": "happy"}, "282": {"wav": "D:/voice-emo/dat/angry\\1001_IOM_ANG_XX_noise_augmented.wav", "label": "angry"}, "283": {"wav": "D:/voice-emo/dat/fear\\1001_TIE_FEA_XX_noise_augmented.wav", "label": "fear"}, "284": {"wav": "D:/voice-emo/dat/neutral\\1001_WSI_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "285": {"wav": "D:/voice-emo/dat/surprise\\DC_su07_stretch_augmented.wav", "label": "surprise"}, "286": {"wav": "D:/voice-emo/dat/fear\\1001_MTI_FEA_XX_noise_augmented.wav", "label": "fear"}, "287": {"wav": "D:/voice-emo/dat/fear\\1001_DFA_FEA_XX_pitch_augmented.wav", "label": "fear"}, "288": {"wav": "D:/voice-emo/dat/neutral\\1001_TIE_NEU_XX_stretch_augmented.wav", "label": "neutral"}, "289": {"wav": "D:/voice-emo/dat/surprise\\DC_su02.wav", "label": "surprise"}, "290": {"wav": "D:/voice-emo/dat/disgust\\1001_MTI_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "291": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_MD_pitch_augmented.wav", "label": "disgust"}, "292": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_HI_pitch_augmented.wav", "label": "happy"}, "293": {"wav": "D:/voice-emo/dat/sad\\1001_TIE_SAD_XX_stretch_augmented.wav", "label": "sad"}}
results/train_with_wav2vec2/1993/valid.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"0": {"wav": "D:/voice-emo/dat/happy\\1001_IWW_HAP_XX_pitch_augmented.wav", "label": "happy"}, "1": {"wav": "D:/voice-emo/dat/happy\\1001_ITH_HAP_XX_noise_augmented.wav", "label": "happy"}, "2": {"wav": "D:/voice-emo/dat/disgust\\1001_ITS_DIS_XX_pitch_augmented.wav", "label": "disgust"}, "3": {"wav": "D:/voice-emo/dat/happy\\1001_DFA_HAP_XX_noise_augmented.wav", "label": "happy"}, "4": {"wav": "D:/voice-emo/dat/disgust\\1001_IWL_DIS_XX.wav", "label": "disgust"}, "5": {"wav": "D:/voice-emo/dat/disgust\\1001_ITH_DIS_XX.wav", "label": "disgust"}, "6": {"wav": "D:/voice-emo/dat/neutral\\1001_DFA_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "7": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_MD.wav", "label": "happy"}, "8": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_HI.wav", "label": "angry"}, "9": {"wav": "D:/voice-emo/dat/happy\\1001_TSI_HAP_XX_noise_augmented.wav", "label": "happy"}, "10": {"wav": "D:/voice-emo/dat/sad\\1001_DFA_SAD_XX_stretch_augmented.wav", "label": "sad"}, "11": {"wav": "D:/voice-emo/dat/neutral\\1001_IOM_NEU_XX_noise_augmented.wav", "label": "neutral"}, "12": {"wav": "D:/voice-emo/dat/sad\\1001_IWW_SAD_XX_noise_augmented.wav", "label": "sad"}, "13": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_LO.wav", "label": "angry"}, "14": {"wav": "D:/voice-emo/dat/happy\\1001_IWL_HAP_XX_pitch_augmented.wav", "label": "happy"}, "15": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_MD_pitch_augmented.wav", "label": "sad"}, "16": {"wav": "D:/voice-emo/dat/angry\\1001_TIE_ANG_XX_noise_augmented.wav", "label": "angry"}, "17": {"wav": "D:/voice-emo/dat/surprise\\DC_su03_pitch_augmented.wav", "label": "surprise"}, "18": {"wav": "D:/voice-emo/dat/neutral\\1001_ITS_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "19": {"wav": "D:/voice-emo/dat/fear\\1001_IOM_FEA_XX_stretch_augmented.wav", "label": "fear"}, "20": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_HI_noise_augmented.wav", "label": "happy"}, "21": {"wav": "D:/voice-emo/dat/neutral\\1001_TSI_NEU_XX_pitch_augmented.wav", "label": "neutral"}, "22": {"wav": "D:/voice-emo/dat/sad\\1001_IEO_SAD_MD.wav", "label": "sad"}, "23": {"wav": "D:/voice-emo/dat/fear\\1001_MTI_FEA_XX_stretch_augmented.wav", "label": "fear"}, "24": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_MD.wav", "label": "angry"}, "25": {"wav": "D:/voice-emo/dat/surprise\\DC_su05.wav", "label": "surprise"}, "26": {"wav": "D:/voice-emo/dat/fear\\1001_IEO_FEA_MD_stretch_augmented.wav", "label": "fear"}, "27": {"wav": "D:/voice-emo/dat/happy\\1001_IEO_HAP_MD_pitch_augmented.wav", "label": "happy"}, "28": {"wav": "D:/voice-emo/dat/happy\\1001_TSI_HAP_XX_stretch_augmented.wav", "label": "happy"}, "29": {"wav": "D:/voice-emo/dat/angry\\1001_IWL_ANG_XX_pitch_augmented.wav", "label": "angry"}, "30": {"wav": "D:/voice-emo/dat/disgust\\1001_IEO_DIS_HI.wav", "label": "disgust"}, "31": {"wav": "D:/voice-emo/dat/happy\\1001_ITH_HAP_XX_pitch_augmented.wav", "label": "happy"}, "32": {"wav": "D:/voice-emo/dat/happy\\1001_IOM_HAP_XX.wav", "label": "happy"}, "33": {"wav": "D:/voice-emo/dat/angry\\1001_IEO_ANG_HI_stretch_augmented.wav", "label": "angry"}, "34": {"wav": "D:/voice-emo/dat/happy\\1001_IWL_HAP_XX_stretch_augmented.wav", "label": "happy"}, "35": {"wav": "D:/voice-emo/dat/neutral\\1001_MTI_NEU_XX_noise_augmented.wav", "label": "neutral"}}
train_with_wav2vec.py ADDED
@@ -0,0 +1,302 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import sys
3
+ import logging
4
+ import speechbrain as sb
5
+ from hyperpyyaml import load_hyperpyyaml
6
+ import json
7
+ import random
8
+ import torch
9
+ from sklearn.preprocessing import LabelEncoder
10
+
11
+ # Check if GPU is available
12
+ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
13
+ print(f"Using device: {device}")
14
+
15
+ logger = logging.getLogger(__name__)
16
+ SAMPLERATE = 16000
17
+
18
+ def prepare_data(data_original, save_json_train, save_json_valid, save_json_test, split_ratio=[80, 10, 10], seed=12):
19
+ # Setting seeds for reproducible code.
20
+ random.seed(seed)
21
+
22
+ # Check if data preparation has already been done (skip if files exist)
23
+ if skip(save_json_train, save_json_valid, save_json_test):
24
+ logger.info("Preparation completed in previous run, skipping.")
25
+ return
26
+
27
+ # Collect audio files and labels
28
+ wav_list = []
29
+ labels = os.listdir(data_original)
30
+
31
+ for label in labels:
32
+ label_dir = os.path.join(data_original, label)
33
+ if os.path.isdir(label_dir):
34
+ for audio_file in os.listdir(label_dir):
35
+ if audio_file.endswith('.wav'):
36
+ wav_file = os.path.join(label_dir, audio_file)
37
+ if os.path.isfile(wav_file):
38
+ wav_list.append((wav_file, label))
39
+ else:
40
+ logger.warning(f"Skipping invalid audio file: {wav_file}")
41
+
42
+ # Shuffle and split the data
43
+ random.shuffle(wav_list)
44
+ n_total = len(wav_list)
45
+ n_train = n_total * split_ratio[0] // 100
46
+ n_valid = n_total * split_ratio[1] // 100
47
+
48
+ train_set = wav_list[:n_train]
49
+ valid_set = wav_list[n_train:n_train + n_valid]
50
+ test_set = wav_list[n_train + n_valid:]
51
+
52
+ # Create JSON files for train, valid, and test sets
53
+ create_json(train_set, save_json_train)
54
+ create_json(valid_set, save_json_valid)
55
+ create_json(test_set, save_json_test)
56
+
57
+ logger.info(f"Created {save_json_train}, {save_json_valid}, and {save_json_test}")
58
+
59
+
60
+ def create_json(wav_list, json_file):
61
+ json_dict = {}
62
+ for wav_file, label in wav_list:
63
+ signal = sb.dataio.dataio.read_audio(wav_file)
64
+ duration = signal.shape[0] / SAMPLERATE
65
+ uttid = os.path.splitext(os.path.basename(wav_file))[0]
66
+
67
+ json_dict[uttid] = {
68
+ "wav": wav_file,
69
+ "length": duration,
70
+ "label": label,
71
+ }
72
+
73
+ with open(json_file, mode="w") as json_f:
74
+ json.dump(json_dict, json_f, indent=2)
75
+
76
+ logger.info(f"Created {json_file}")
77
+
78
+
79
+ def skip(*filenames):
80
+ for filename in filenames:
81
+ if not os.path.isfile(filename):
82
+ return False
83
+ return True
84
+
85
+
86
+ class EmoIdBrain(sb.Brain):
87
+ def compute_forward(self, batch, stage):
88
+ """Computation pipeline based on an encoder + emotion classifier."""
89
+ batch = batch.to(self.device)
90
+ wavs, lens = batch.sig
91
+
92
+ outputs = self.modules.wav2vec2(wavs, lens)
93
+
94
+ # Apply pooling and MLP layers
95
+ outputs = self.hparams.avg_pool(outputs, lens)
96
+ outputs = outputs.view(outputs.shape[0], -1)
97
+ outputs = self.modules.output_mlp(outputs)
98
+ outputs = self.hparams.log_softmax(outputs)
99
+
100
+ return outputs
101
+
102
+ def compute_objectives(self, predictions, batch, stage):
103
+ emo_encoded_list = []
104
+
105
+ for sample in batch:
106
+ # Check if 'emo_encoded' exists in the sample
107
+ if 'emo_encoded' in sample:
108
+ emo_encoded_list.append(sample['emo_encoded'])
109
+ else:
110
+ # Log a warning and skip this sample if 'emo_encoded' is missing
111
+ logging.warning(f"'emo_encoded' key not found in sample: {sample}")
112
+
113
+ if not emo_encoded_list:
114
+ # If no valid 'emo_encoded' values were found in the batch, raise an error
115
+ raise ValueError("No valid 'emo_encoded' values found in the batch.")
116
+
117
+ # Convert emo_encoded_list to a torch tensor
118
+ emo_encoded = torch.tensor(emo_encoded_list, dtype=torch.long)
119
+
120
+ # Ensure emo_encoded is a tensor
121
+ if not isinstance(emo_encoded, torch.Tensor):
122
+ raise TypeError(f"Unsupported label type encountered: {type(emo_encoded)}")
123
+
124
+ # Perform any necessary operations with emo_encoded here
125
+ loss = self.hparams.compute_cost(predictions, emo_encoded)
126
+
127
+ if stage != sb.Stage.TRAIN:
128
+ self.error_metrics.append(batch.id, predictions, emo_encoded)
129
+
130
+ return loss
131
+
132
+
133
+ def on_stage_start(self, stage, epoch=None):
134
+ """Gets called at the beginning of each epoch."""
135
+ self.loss_metric = sb.utils.metric_stats.MetricStats(metric=sb.nnet.losses.nll_loss)
136
+ if stage != sb.Stage.TRAIN:
137
+ self.error_metrics = self.hparams.error_stats()
138
+
139
+ def on_stage_end(self, stage, stage_loss, epoch=None):
140
+ """Gets called at the end of an epoch."""
141
+ if stage == sb.Stage.TRAIN:
142
+ self.train_loss = stage_loss
143
+ else:
144
+ stats = {
145
+ "loss": stage_loss,
146
+ }
147
+ if self.error_metrics is not None and len(self.error_metrics.scores) > 0:
148
+ # Calculate error rate only if there are scores in the error_metrics
149
+ stats["error_rate"] = self.error_metrics.summarize("average")
150
+ else:
151
+ # Handle case where error_metrics are None or empty
152
+ stats["error_rate"] = float('nan') # Set error_rate to NaN if no scores available
153
+
154
+ if stage == sb.Stage.VALID:
155
+ old_lr, new_lr = self.hparams.lr_annealing(stats["error_rate"])
156
+ sb.nnet.schedulers.update_learning_rate(self.optimizer, new_lr)
157
+ self.hparams.train_logger.log_stats(
158
+ {"Epoch": epoch, "lr": old_lr},
159
+ train_stats={"loss": self.train_loss},
160
+ valid_stats=stats,
161
+ )
162
+ self.checkpointer.save_and_keep_only(meta=stats, min_keys=["error_rate"])
163
+ elif stage == sb.Stage.TEST:
164
+ self.hparams.train_logger.log_stats(
165
+ {"Epoch loaded": self.hparams.epoch_counter.current},
166
+ test_stats=stats,
167
+ )
168
+
169
+ def init_optimizers(self):
170
+ """Initializes the optimizer."""
171
+ self.optimizer = self.hparams.opt_class(self.hparams.model.parameters())
172
+ if self.checkpointer is not None:
173
+ self.checkpointer.add_recoverable("optimizer", self.optimizer)
174
+ self.optimizers_dict = {"model_optimizer": self.optimizer}
175
+
176
+
177
+ def dataio_prep(hparams):
178
+ """Prepares the datasets to be used in the brain class."""
179
+
180
+ # Define the audio processing pipeline
181
+ @sb.utils.data_pipeline.takes("wav")
182
+ @sb.utils.data_pipeline.provides("sig")
183
+ def audio_pipeline(wav):
184
+ """Load the signal from a WAV file."""
185
+ sig = sb.dataio.dataio.read_audio(wav)
186
+ return sig
187
+
188
+ # Initialize the label encoder
189
+ label_encoder = sb.dataio.encoder.CategoricalEncoder()
190
+ label_encoder.add_unk()
191
+
192
+ label_to_index = {
193
+ 'angry': 0,
194
+ 'happy': 1,
195
+ 'neutral': 2,
196
+ 'sad': 3,
197
+ 'surprise': 4,
198
+ 'disgust': 5,
199
+ 'fear': 6
200
+ }
201
+
202
+ @sb.utils.data_pipeline.takes("label")
203
+ @sb.utils.data_pipeline.provides("label", "emo_encoded")
204
+ def label_pipeline(label):
205
+ """Encode the emotion label."""
206
+ if label in label_to_index:
207
+ emo_encoded = label_to_index[label]
208
+ else:
209
+ raise ValueError(f"Unknown label encountered: {label}")
210
+
211
+ yield label, torch.tensor(emo_encoded, dtype=torch.long)
212
+
213
+ # Define datasets dictionary
214
+ datasets = {}
215
+ data_info = {
216
+ "train": hparams["train_annotation"],
217
+ "valid": hparams["valid_annotation"],
218
+ "test": hparams["test_annotation"],
219
+ }
220
+
221
+ # Load datasets and apply pipelines
222
+ for dataset_name, json_path in data_info.items():
223
+ datasets[dataset_name] = sb.dataio.dataset.DynamicItemDataset.from_json(
224
+ json_path=json_path,
225
+ replacements={"data_root": hparams["data_original"]},
226
+ dynamic_items=[audio_pipeline, label_pipeline],
227
+ output_keys=["id", "sig", "label", "emo_encoded"],
228
+ )
229
+
230
+ lab_enc_file = os.path.join(hparams["save_folder"], "label_encoder.txt")
231
+ label_encoder.load_or_create(
232
+ path=lab_enc_file,
233
+ from_didatasets=[datasets["train"]],
234
+ output_key="label",
235
+ )
236
+
237
+ return datasets
238
+
239
+
240
+ if __name__ == "__main__":
241
+ hparams_file, run_opts, overrides = sb.parse_arguments(sys.argv[1:])
242
+ sb.utils.distributed.ddp_init_group(run_opts)
243
+
244
+ try:
245
+ with open(hparams_file) as fin:
246
+ hparams = load_hyperpyyaml(fin, overrides)
247
+ data_original = hparams.get("data_original")
248
+ if data_original is not None:
249
+ data_original = os.path.normpath(data_original)
250
+ if not os.path.exists(data_original):
251
+ raise ValueError(f"data_original path '{data_original}' does not exist.")
252
+ else:
253
+ raise ValueError("data_original path is not specified in the YAML configuration.")
254
+
255
+ except Exception as e:
256
+ print("Error occurred", e)
257
+ sys.exit(1)
258
+
259
+ sb.create_experiment_directory(
260
+ experiment_directory=hparams["output_folder"],
261
+ hyperparams_to_save=hparams_file,
262
+ overrides=overrides,
263
+ )
264
+
265
+ if not hparams["skip_prep"]:
266
+ prepare_kwargs = {
267
+ "data_original": hparams["data_original"],
268
+ "save_json_train": hparams["train_annotation"],
269
+ "save_json_valid": hparams["valid_annotation"],
270
+ "save_json_test": hparams["test_annotation"],
271
+ "split_ratio": hparams["split_ratio"],
272
+ "seed": hparams["seed"],
273
+ }
274
+ sb.utils.distributed.run_on_main(prepare_data, kwargs=prepare_kwargs)
275
+
276
+ datasets = dataio_prep(hparams)
277
+
278
+ hparams["wav2vec2"] = hparams["wav2vec2"].to(device=run_opts["device"])
279
+ if not hparams["freeze_wav2vec2"] and hparams["freeze_wav2vec2_conv"]:
280
+ hparams["wav2vec2"].model.feature_extractor._freeze_parameters()
281
+
282
+ emo_id_brain = EmoIdBrain(
283
+ modules=hparams["modules"],
284
+ opt_class=hparams["opt_class"],
285
+ hparams=hparams,
286
+ run_opts=run_opts,
287
+ checkpointer=hparams["checkpointer"],
288
+ )
289
+
290
+ emo_id_brain.fit(
291
+ epoch_counter=emo_id_brain.hparams.epoch_counter,
292
+ train_set=datasets["train"],
293
+ valid_set=datasets["valid"],
294
+ train_loader_kwargs=hparams["dataloader_options"],
295
+ valid_loader_kwargs=hparams["dataloader_options"],
296
+ )
297
+
298
+ test_stats = emo_id_brain.evaluate(
299
+ test_set=datasets["test"],
300
+ min_key="error_rate",
301
+ test_loader_kwargs=hparams["dataloader_options"],
302
+ )