TTS/hparams.py at main · DS-AEON/TTS · 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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
from text import symbols
import argparse

defaults = {
    ################################
    # Experiment Parameters        #
    ################################
    'epochs':500,
    'iters_per_checkpoint':10000,
    'seed':1234,
    'dynamic_loss_scaling':True,
    'fp16_run':False,
    'distributed_run':False,
    'dist_backend':"nccl",
    'dist_url':"tcp://localhost:54321",
    'cudnn_enabled':True,
    'cudnn_benchmark':False,
    'ignore_layers':"['embedding.weight']",

    ################################
    # Data Parameters             #
    ################################
    'load_mel_from_disk':False,
    'training_files':'filelists/train_filelist.txt',
    'validation_files':'filelists/val_filelist.txt',
    'text_cleaners':"['korean_cleaners']",  ##한글을 사용하려면 "['korean_cleaners']"   -> str 형태로 입력해야 함
                                            ##영어를 사용하려면 "['english_cleaners']"
    ################################
    # Audio Parameters             #
    ################################
    'max_wav_value':32768.0,
    'sampling_rate':22050,
    'filter_length':1024,
    'hop_length':256,
    'win_length':1024,
    'n_mel_channels':80,
    'mel_fmin':0.0,
    'mel_fmax':8000.0,

    ################################
    # Model Parameters             #
    ################################
    'n_symbols':len(symbols),
    'symbols_embedding_dim':512,

    # Encoder parameters
    'encoder_kernel_size':5,
    'encoder_n_convolutions':3,
    'encoder_embedding_dim':512,

    # Decoder parameters
    'n_frames_per_step':1,  # currently only 1 is supported
    'decoder_rnn_dim':1024,
    'prenet_dim':256,
    'max_decoder_steps':1000,
    'gate_threshold':0.5,
    'p_attention_dropout':0.1,
    'p_decoder_dropout':0.1,

    # Attention parameters
    'attention_rnn_dim':1024,
    'attention_dim':128,

    # Location Layer parameters
    'attention_location_n_filters':32,
    'attention_location_kernel_size':31,

    # Mel-post processing network parameters
    'postnet_embedding_dim':512,
    'postnet_kernel_size':5,
    'postnet_n_convolutions':5,

    ################################
    # Optimization Hyperparameters #
    ################################
    'use_saved_learning_rate':False,
    'learning_rate':1e-3,
    'weight_decay':1e-6,
    'grad_clip_thresh':1.0,
    'batch_size':32,
    'gradient_accumulation_steps':2,
    'mask_padding':True  # set model's padded outputs to padded values
}

def get_hparams(args, parser):
    arg_groups = {}
    for group in parser._action_groups:
        group_dict = {a.dest: getattr(args, a.dest, None) for a in group._group_actions}
        arg_groups[group.title] = argparse.Namespace(**group_dict)
    ###################
    hparams = arg_groups['hparams']
    hparams.text_cleaners = eval(hparams.text_cleaners)
    hparams.ignore_layers = eval(hparams.ignore_layers)

    return hparams

def add_hparams(parser):
    """Create model hyperparameters. Parse nondefault from given string."""

    hparams_group = parser.add_argument_group('hparams')
    for key, default in defaults.items():
        hparams_group.add_argument("--{}".format(key), type=type(default), default=default)


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    hparams_group = parser.add_argument_group('hparams')
    for key, default in defaults.items():
        hparams_group.add_argument("--{}".format(key), type=type(default), default=default)
    args = parser.parse_args()

    arg_groups = {}
    for group in parser._action_groups:
        group_dict = {a.dest: getattr(args, a.dest, None) for a in group._group_actions}
        arg_groups[group.title] = argparse.Namespace(**group_dict)
    ###################
    hparams = arg_groups['hparams']

    print(hparams)
    print(type(hparams.text_cleaners))
    print(type(eval(hparams.text_cleaners)))
    print("############################")
    print(args)