Spaces:
Runtime error
Runtime error
update
Browse files- checkpoints/fs2_exp/config.yaml +1 -1
- checkpoints/fs2_exp/{model_ckpt_steps_98000.ckpt → model_ckpt_steps_160000.ckpt} +1 -1
- checkpoints/fs_pitch/config.yaml +216 -0
- checkpoints/fs_pitch/model_ckpt_steps_160000.ckpt +3 -0
- checkpoints/ps_normal_exp/config.yaml +1 -1
- checkpoints/ps_normal_exp/{model_ckpt_steps_278000.ckpt → model_ckpt_steps_406000.ckpt} +1 -1
- checkpoints/ps_small_exp/config.yaml +1 -1
- checkpoints/ps_small_exp/model_ckpt_steps_410000.ckpt +1 -1
checkpoints/fs2_exp/config.yaml
CHANGED
|
@@ -64,7 +64,7 @@ encoder_K: 8
|
|
| 64 |
encoder_type: fft
|
| 65 |
endless_ds: true
|
| 66 |
eval_max_batches: -1
|
| 67 |
-
f0_max:
|
| 68 |
f0_min: 80
|
| 69 |
ffn_act: gelu
|
| 70 |
ffn_hidden_size: 1024
|
|
|
|
| 64 |
encoder_type: fft
|
| 65 |
endless_ds: true
|
| 66 |
eval_max_batches: -1
|
| 67 |
+
f0_max: 600
|
| 68 |
f0_min: 80
|
| 69 |
ffn_act: gelu
|
| 70 |
ffn_hidden_size: 1024
|
checkpoints/fs2_exp/{model_ckpt_steps_98000.ckpt → model_ckpt_steps_160000.ckpt}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 108423039
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f73b9e52933d3e2bcf6591ac375ab25dc3bdc4b03e1ae5058fc003fa0dd721e
|
| 3 |
size 108423039
|
checkpoints/fs_pitch/config.yaml
ADDED
|
@@ -0,0 +1,216 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
accumulate_grad_batches: 1
|
| 2 |
+
amp: false
|
| 3 |
+
audio_num_mel_bins: 80
|
| 4 |
+
audio_sample_rate: 22050
|
| 5 |
+
base_config:
|
| 6 |
+
- egs/egs_bases/tts/fs.yaml
|
| 7 |
+
- ./base_text2mel.yaml
|
| 8 |
+
binarization_args:
|
| 9 |
+
min_sil_duration: 0.1
|
| 10 |
+
shuffle: false
|
| 11 |
+
test_range:
|
| 12 |
+
- 0
|
| 13 |
+
- 523
|
| 14 |
+
train_range:
|
| 15 |
+
- 871
|
| 16 |
+
- -1
|
| 17 |
+
trim_eos_bos: false
|
| 18 |
+
valid_range:
|
| 19 |
+
- 523
|
| 20 |
+
- 871
|
| 21 |
+
with_align: true
|
| 22 |
+
with_f0: true
|
| 23 |
+
with_f0cwt: false
|
| 24 |
+
with_linear: false
|
| 25 |
+
with_spk_embed: false
|
| 26 |
+
with_wav: false
|
| 27 |
+
binarizer_cls: data_gen.tts.base_binarizer.BaseBinarizer
|
| 28 |
+
binary_data_dir: data/binary/ljspeech
|
| 29 |
+
check_val_every_n_epoch: 10
|
| 30 |
+
clip_grad_norm: 1
|
| 31 |
+
clip_grad_value: 0
|
| 32 |
+
conv_use_pos: false
|
| 33 |
+
debug: false
|
| 34 |
+
dec_dilations:
|
| 35 |
+
- 1
|
| 36 |
+
- 1
|
| 37 |
+
- 1
|
| 38 |
+
- 1
|
| 39 |
+
dec_ffn_kernel_size: 9
|
| 40 |
+
dec_inp_add_noise: false
|
| 41 |
+
dec_kernel_size: 5
|
| 42 |
+
dec_layers: 4
|
| 43 |
+
dec_post_net_kernel: 3
|
| 44 |
+
decoder_rnn_dim: 0
|
| 45 |
+
decoder_type: conv
|
| 46 |
+
dropout: 0.0
|
| 47 |
+
ds_workers: 2
|
| 48 |
+
dur_predictor_kernel: 3
|
| 49 |
+
dur_predictor_layers: 2
|
| 50 |
+
enc_dec_norm: ln
|
| 51 |
+
enc_dilations:
|
| 52 |
+
- 1
|
| 53 |
+
- 1
|
| 54 |
+
- 1
|
| 55 |
+
- 1
|
| 56 |
+
enc_ffn_kernel_size: 9
|
| 57 |
+
enc_kernel_size: 5
|
| 58 |
+
enc_layers: 4
|
| 59 |
+
enc_post_net_kernel: 3
|
| 60 |
+
enc_pre_ln: true
|
| 61 |
+
enc_prenet: true
|
| 62 |
+
encoder_K: 8
|
| 63 |
+
encoder_type: rel_fft
|
| 64 |
+
endless_ds: true
|
| 65 |
+
eval_max_batches: -1
|
| 66 |
+
f0_max: 600
|
| 67 |
+
f0_min: 80
|
| 68 |
+
ffn_act: gelu
|
| 69 |
+
ffn_hidden_size: 1024
|
| 70 |
+
fft_size: 1024
|
| 71 |
+
fmax: 7600
|
| 72 |
+
fmin: 80
|
| 73 |
+
frames_multiple: 1
|
| 74 |
+
gen_dir_name: ''
|
| 75 |
+
griffin_lim_iters: 30
|
| 76 |
+
hidden_size: 256
|
| 77 |
+
hop_size: 256
|
| 78 |
+
infer: false
|
| 79 |
+
lambda_commit: 0.25
|
| 80 |
+
lambda_energy: 0.1
|
| 81 |
+
lambda_f0: 1.0
|
| 82 |
+
lambda_ph_dur: 0.1
|
| 83 |
+
lambda_sent_dur: 1.0
|
| 84 |
+
lambda_uv: 1.0
|
| 85 |
+
lambda_word_dur: 1.0
|
| 86 |
+
layers_in_block: 2
|
| 87 |
+
load_ckpt: ''
|
| 88 |
+
loud_norm: false
|
| 89 |
+
lr: 0.0005
|
| 90 |
+
max_epochs: 1000
|
| 91 |
+
max_frames: 1548
|
| 92 |
+
max_input_tokens: 1550
|
| 93 |
+
max_sentences: 128
|
| 94 |
+
max_tokens: 40000
|
| 95 |
+
max_updates: 160000
|
| 96 |
+
max_valid_sentences: 1
|
| 97 |
+
max_valid_tokens: 60000
|
| 98 |
+
mel_losses: l1:0.5|ssim:0.5
|
| 99 |
+
mel_vmax: 1.5
|
| 100 |
+
mel_vmin: -6
|
| 101 |
+
min_frames: 0
|
| 102 |
+
num_ckpt_keep: 3
|
| 103 |
+
num_heads: 2
|
| 104 |
+
num_sanity_val_steps: 5
|
| 105 |
+
num_spk: 1
|
| 106 |
+
num_valid_plots: 10
|
| 107 |
+
optimizer_adam_beta1: 0.9
|
| 108 |
+
optimizer_adam_beta2: 0.98
|
| 109 |
+
out_wav_norm: false
|
| 110 |
+
pitch_extractor: parselmouth
|
| 111 |
+
pitch_key: pitch
|
| 112 |
+
pitch_type: frame
|
| 113 |
+
predictor_dropout: 0.5
|
| 114 |
+
predictor_grad: 0.1
|
| 115 |
+
predictor_hidden: -1
|
| 116 |
+
predictor_kernel: 5
|
| 117 |
+
predictor_layers: 2
|
| 118 |
+
preprocess_args:
|
| 119 |
+
add_eos_bos: true
|
| 120 |
+
mfa_group_shuffle: false
|
| 121 |
+
mfa_offset: 0.02
|
| 122 |
+
nsample_per_mfa_group: 1000
|
| 123 |
+
reset_phone_dict: true
|
| 124 |
+
reset_word_dict: true
|
| 125 |
+
save_sil_mask: true
|
| 126 |
+
txt_processor: en
|
| 127 |
+
use_mfa: true
|
| 128 |
+
vad_max_silence_length: 12
|
| 129 |
+
wav_processors: []
|
| 130 |
+
with_phsep: true
|
| 131 |
+
preprocess_cls: egs.datasets.audio.lj.preprocess.LJPreprocess
|
| 132 |
+
print_nan_grads: false
|
| 133 |
+
processed_data_dir: data/processed/ljspeech
|
| 134 |
+
profile_infer: false
|
| 135 |
+
raw_data_dir: data/raw/LJSpeech-1.1
|
| 136 |
+
ref_norm_layer: bn
|
| 137 |
+
rename_tmux: true
|
| 138 |
+
resume_from_checkpoint: 0
|
| 139 |
+
save_best: false
|
| 140 |
+
save_codes:
|
| 141 |
+
- tasks
|
| 142 |
+
- modules
|
| 143 |
+
- egs
|
| 144 |
+
save_f0: false
|
| 145 |
+
save_gt: true
|
| 146 |
+
scheduler: warmup
|
| 147 |
+
seed: 1234
|
| 148 |
+
sort_by_len: true
|
| 149 |
+
task_cls: tasks.tts.fs.FastSpeechTask
|
| 150 |
+
tb_log_interval: 100
|
| 151 |
+
test_ids:
|
| 152 |
+
- 0
|
| 153 |
+
- 1
|
| 154 |
+
- 2
|
| 155 |
+
- 3
|
| 156 |
+
- 4
|
| 157 |
+
- 5
|
| 158 |
+
- 6
|
| 159 |
+
- 7
|
| 160 |
+
- 8
|
| 161 |
+
- 9
|
| 162 |
+
- 10
|
| 163 |
+
- 11
|
| 164 |
+
- 12
|
| 165 |
+
- 13
|
| 166 |
+
- 14
|
| 167 |
+
- 15
|
| 168 |
+
- 16
|
| 169 |
+
- 17
|
| 170 |
+
- 18
|
| 171 |
+
- 19
|
| 172 |
+
- 68
|
| 173 |
+
- 70
|
| 174 |
+
- 74
|
| 175 |
+
- 87
|
| 176 |
+
- 110
|
| 177 |
+
- 172
|
| 178 |
+
- 190
|
| 179 |
+
- 215
|
| 180 |
+
- 231
|
| 181 |
+
- 294
|
| 182 |
+
- 316
|
| 183 |
+
- 324
|
| 184 |
+
- 402
|
| 185 |
+
- 422
|
| 186 |
+
- 485
|
| 187 |
+
- 500
|
| 188 |
+
- 505
|
| 189 |
+
- 508
|
| 190 |
+
- 509
|
| 191 |
+
- 519
|
| 192 |
+
test_input_yaml: ''
|
| 193 |
+
test_num: 100
|
| 194 |
+
test_set_name: test
|
| 195 |
+
train_set_name: train
|
| 196 |
+
train_sets: ''
|
| 197 |
+
use_gt_dur: false
|
| 198 |
+
use_gt_f0: false
|
| 199 |
+
use_pitch_embed: true
|
| 200 |
+
use_pos_embed: true
|
| 201 |
+
use_spk_embed: false
|
| 202 |
+
use_spk_id: false
|
| 203 |
+
use_uv: true
|
| 204 |
+
use_word_input: false
|
| 205 |
+
val_check_interval: 2000
|
| 206 |
+
valid_infer_interval: 10000
|
| 207 |
+
valid_monitor_key: val_loss
|
| 208 |
+
valid_monitor_mode: min
|
| 209 |
+
valid_set_name: valid
|
| 210 |
+
vocoder: HifiGAN
|
| 211 |
+
vocoder_ckpt: checkpoints/hifi_lj
|
| 212 |
+
warmup_updates: 4000
|
| 213 |
+
weight_decay: 0
|
| 214 |
+
win_size: 1024
|
| 215 |
+
word_dict_size: 10000
|
| 216 |
+
work_dir: checkpoints/1402_fs2_lj_1
|
checkpoints/fs_pitch/model_ckpt_steps_160000.ckpt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f3bd822cd2a1d976e7edc0d6bf84f13d06f01265efdcac88880bae5ae3e7140
|
| 3 |
+
size 85090681
|
checkpoints/ps_normal_exp/config.yaml
CHANGED
|
@@ -65,7 +65,7 @@ encoder_K: 8
|
|
| 65 |
encoder_type: rel_fft
|
| 66 |
endless_ds: true
|
| 67 |
eval_max_batches: -1
|
| 68 |
-
f0_max:
|
| 69 |
f0_min: 80
|
| 70 |
ffn_act: gelu
|
| 71 |
ffn_hidden_size: 768
|
|
|
|
| 65 |
encoder_type: rel_fft
|
| 66 |
endless_ds: true
|
| 67 |
eval_max_batches: -1
|
| 68 |
+
f0_max: 600
|
| 69 |
f0_min: 80
|
| 70 |
ffn_act: gelu
|
| 71 |
ffn_hidden_size: 768
|
checkpoints/ps_normal_exp/{model_ckpt_steps_278000.ckpt → model_ckpt_steps_406000.ckpt}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 104081102
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:045ea7ba3df9eae43347e8cd59c872f501b9cfd4c0b22fe1d610cdc8ab0b8aee
|
| 3 |
size 104081102
|
checkpoints/ps_small_exp/config.yaml
CHANGED
|
@@ -65,7 +65,7 @@ encoder_K: 8
|
|
| 65 |
encoder_type: rel_fft
|
| 66 |
endless_ds: true
|
| 67 |
eval_max_batches: -1
|
| 68 |
-
f0_max:
|
| 69 |
f0_min: 80
|
| 70 |
ffn_act: gelu
|
| 71 |
ffn_hidden_size: 512
|
|
|
|
| 65 |
encoder_type: rel_fft
|
| 66 |
endless_ds: true
|
| 67 |
eval_max_batches: -1
|
| 68 |
+
f0_max: 600
|
| 69 |
f0_min: 80
|
| 70 |
ffn_act: gelu
|
| 71 |
ffn_hidden_size: 512
|
checkpoints/ps_small_exp/model_ckpt_steps_410000.ckpt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 32754716
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:760eb5c55d29327f3b4ccdff425c93bafc1d30eae90c218bf81a3a2afbe7f4be
|
| 3 |
size 32754716
|