End-to-end-ASR-Pytorch/data/preprocess_timit.py at master · renrenzsbbb/End-to-end-ASR-Pytorch · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
import sys
sys.path.insert(0, '..')
from src.preprocess import extract_feature,encode_target
from joblib import Parallel, delayed
import argparse
import os
from pathlib import Path
from tqdm import tqdm
import pickle

def boolean_string(s):
    if s not in ['False', 'True']:
        raise ValueError('Not a valid boolean string')
    return s == 'True'

parser = argparse.ArgumentParser(description='Preprocess program for TIMIT dataset.')
parser.add_argument('--data_path', type=str, help='Path to raw TIMIT dataset')
parser.add_argument('--feature_type', default='mfcc', type=str, help='Feature type ( mfcc / fbank )', required=False)
parser.add_argument('--feature_dim', default=13, type=int, help='Dimension of feature', required=False)
parser.add_argument('--apply_delta', default=True, type=boolean_string, help='Append Delta', required=False)
parser.add_argument('--apply_delta_delta', default=True, type=boolean_string, help='Append Delta Delta', required=False)
parser.add_argument('--apply_cmvn', default=True, type=boolean_string, help='Apply CMVN on feature', required=False)
parser.add_argument('--output_path', default='.', type=str, help='Path to store output', required=False)
parser.add_argument('--n_jobs', default=-1, type=int, help='Number of jobs used for feature extraction', required=False)
parser.add_argument('--target', default='phoneme', type=str, help='Learning target ( phoneme / char / subword / word )', required=False)
parser.add_argument('--n_tokens', default=1000, type=int, help='Vocabulary size of target', required=False)
paras = parser.parse_args()

def read_text(file,target):
    labels = []
    if target == 'phoneme':
        with open(file.replace('.wav','.phn'),'r') as f:
            for line in f:
                labels.append(line.replace('\n','').split(' ')[-1])
    elif target in ['char','subword','word']:
        with open(file.replace('.wav','.wrd'),'r') as f:
            for line in f:
                labels.append(line.replace('\n','').split(' ')[-1])
        if target =='char':
            labels = [c for c in ' '.join(labels)]
    else:
        raise ValueError('Unsupported target: '+target)
    return labels


# Process training data
print('')
print('Preprocessing training data...',end='')
todo = list(Path(os.path.join(paras.data_path,'train')).rglob("*.[wW][aA][vV]"))
print(len(todo),'audio files found in training set (should be 4620)')


print('Extracting acoustic feature...',flush=True)
tr_x = Parallel(n_jobs=paras.n_jobs)(delayed(extract_feature)(str(file),feature=paras.feature_type,dim=paras.feature_dim,cmvn=paras.apply_cmvn,delta=paras.apply_delta,delta_delta=paras.apply_delta_delta) for file in tqdm(todo))
print('Encoding training target...',flush=True)
tr_y = Parallel(n_jobs=paras.n_jobs)(delayed(read_text)(str(file),target=paras.target)                               for file in tqdm(todo))
tr_y, encode_table = encode_target(tr_y,table=None,mode=paras.target,max_idx=paras.n_tokens)

dim = paras.feature_dim*(1+paras.apply_delta+paras.apply_delta_delta)
output_dir = os.path.join(paras.output_path,'_'.join(['timit',str(paras.feature_type)+str(dim),str(paras.target)+str(len(encode_table))]))
print('Saving training data to',output_dir)
if not os.path.exists(output_dir):os.makedirs(output_dir)
with open(os.path.join(output_dir,"train_x.pkl"), "wb") as fp:
    pickle.dump(tr_x, fp)
del tr_x
with open(os.path.join(output_dir,"train_y.pkl"), "wb") as fp:
    pickle.dump(tr_y, fp)
del tr_y
with open(os.path.join(output_dir,"mapping.pkl"), "wb") as fp:
    pickle.dump(encode_table, fp)


# Process testing data
print('Preprocessing testing data...',end='')
todo = list(Path(os.path.join(paras.data_path,'test')).rglob("*.[wW][aA][vV]"))
print(len(todo),'audio files found in test set (should be 1680)')

print('Extracting acoustic feature...',flush=True)
tt_x = Parallel(n_jobs=paras.n_jobs)(delayed(extract_feature)(str(file),feature=paras.feature_type,dim=paras.feature_dim,cmvn=paras.apply_cmvn,delta=paras.apply_delta,delta_delta=paras.apply_delta_delta)                               for file in tqdm(todo))
print('Encoding testing target...',flush=True)
tt_y = Parallel(n_jobs=paras.n_jobs)(delayed(read_text)(str(file),target=paras.target)                               for file in tqdm(todo))
tt_y, encode_table = encode_target(tt_y,table=encode_table,mode=paras.target,max_idx=paras.n_tokens)


print('Saving testing data to',output_dir)
if not os.path.exists(output_dir):os.makedirs(output_dir)
with open(os.path.join(output_dir,"test_x.pkl"), "wb") as fp:
    pickle.dump(tt_x, fp)
del tt_x
with open(os.path.join(output_dir,"test_y.pkl"), "wb") as fp:
    pickle.dump(tt_y, fp)
del tt_y
print('All done, exit.')