ljspeech_transformer_tts_aept/config.yml

additional_config: conf/tts_aept.v1.yaml
allow_cache: true
batch_size: 100
cleaner: tacotron
config: exp/tts_aept_phn_tacotron_lr8e-04_wu4k_r2_eos_bs128_checkpoint-86000steps/original_config.yaml
dev_dumpdir: dump/dev/norm
dev_text: data/dev/text
distributed: false
eval_interval_steps: 100
fft_size: 1024
fmax: 7600
fmin: 80
format: hdf5
freeze-mods:
- decoder
- feat_out
- prob_out
- postnet
g2p: g2p_en
global_gain_scale: 0.95
grad_norm: 1.0
hop_size: 256
inference:
  maxlenratio: 2.0
  minlenratio: 0.0
  threshold: 0.5
init-mods:
- decoder
- feat_out
- prob_out
- postnet
init_checkpoint: exp/tts_aept_phn_tacotron_lr8e-04_wu4k_r2_eos_bs128_checkpoint-86000steps/original_checkpoint-86000steps.pkl
log_interval_steps: 10
model_params:
  adim: 384
  aheads: 4
  decoder_concat_after: false
  decoder_normalize_before: false
  decoder_reduction_factor: 2
  dlayers: 6
  dprenet_layers: 2
  dprenet_units: 256
  dunits: 1536
  elayers: 6
  encoder_concat_after: false
  encoder_normalize_before: true
  eunits: 1536
  idim: 80
  odim: 80
  postnet_chans: 256
  postnet_filts: 5
  postnet_layers: 5
  use_batch_norm: true
model_type: VTN
non_linguistic_symbols: none
num_mels: 80
num_save_intermediate_results: 4
num_workers: 2
optimizer_params:
  lr: 0.0008
optimizer_type: Adam
outdir: exp/tts_aept_phn_tacotron_lr8e-04_wu4k_r2_eos_bs128_checkpoint-86000steps
pin_memory: true
pretrain: ''
rank: 0
resume: ''
sampling_rate: 16000
save_interval_steps: 1000
scheduler: warmuplr
scheduler_params:
  warmup_steps: 4000
seq2seq_loss_params:
  bce_pos_weight: 10.0
src_dev_dumpdir: dump/dev/norm
src_feat_type: mel
src_train_dumpdir: dump/train_no_dev/norm
stats:
  mean: !!python/object/apply:numpy.core.multiarray._reconstruct
    args:
    - &id001 !!python/name:numpy.ndarray ''
    - !!python/tuple
      - 0
    - !!binary |
      Yg==
    state: !!python/tuple
    - 1
    - !!python/tuple
      - 80
    - &id002 !!python/object/apply:numpy.dtype
      args:
      - f4
      - false
      - true
      state: !!python/tuple
      - 3
      - <
      - null
      - null
      - null
      - -1
      - -1
      - 0
    - false
    - !!binary |
      uCcJwJgR6b8mW8i/AqbEvxauxr+NKse/N6jHvytBx7/vdse/HKm6vwDiu79Uqb2/p0TDvxdgy78k
      kte/Xd7dv+GN4L/PK+e/Mejvv2Uc+b8aqgHAqXwGwPwDCMCsKQnAc2UKwMtPDMCAfQzAZw0RwABs
      EsDMIhXA8fgXwJSCGsAm8xvAlWAdwDikHcBZyRrA8G0YwAXmFcCKGhXA9iYUwCbUEsBLwxHA3QcU
      wJIXF8C7+BrAKSQewBa5IMARwyHAJAwhwB6dIMDxHh/AvbUdwM1mG8CfQRnAvOUYwFuOGsD58hzA
      43sewA1aH8CKqx7Aw8sdwJYFHsCw9x7A8SAhwOs6I8Bf7CTAx5UmwBgVKMAZYCnABEMtwMzxM8BD
      qTvAJm9BwOxVRcAt2EXA1eJEwKxhQcCR7jvAbdQ4wE6NRMA=
  scale: !!python/object/apply:numpy.core.multiarray._reconstruct
    args:
    - *id001
    - !!python/tuple
      - 0
    - !!binary |
      Yg==
    state: !!python/tuple
    - 1
    - !!python/tuple
      - 80
    - *id002
    - false
    - !!binary |
      DP3YPtefJj/bkUY/XPhDP5TSNz+WMjY/rUpGP3CyTD9GiE0/VthAPzDZQj8p0EI/5f1EP764QD94
      ZEA/ReZCPwz8Qz95UUI/NkNAPzNtPz9gUUU/bHZEP0JDQz/oekU/+XtFP+rMQz8MWkA/Vo1AP5eL
      Pz/urDs/lF4+PzP4QD8Of0E/dX5AP9byPj/JPzw/8Iw8P2qIPj+1tD8/o6BAP/UWQT9askA/13k+
      P9vaOj9uMjg/Nws3P9FUNz+lWjg/a+c7P7LtQD8AIkM/ED1FP91FRT9RiEc/N5NKP9GBRz8qz0E/
      0lo8Pz+DOT9Kzzo//qo8PwrzPT/e8EA/4PtEP74YSD+z00g/MLlIP++vSj/ZnE4/nvVSPzduVz+G
      VF4/q8llP9/lbT+d4HE/+wxvPz0JbT+6Um0/kXJuP580bz8=
token_list: exp/phn_tacotron_lr8e-04_wu4k_r2_eos_bs128/tokens.txt
token_type: phn
train_dumpdir: dump/train_no_dev/norm
train_max_steps: 50000
train_text: data/train_no_dev/text
trg_dev_dumpdir: dump/dev/norm
trg_feat_type: mel
trg_stats: exp/tts_aept_phn_tacotron_lr8e-04_wu4k_r2_eos_bs128_checkpoint-86000steps/stats.h5
trg_train_dumpdir: dump/train_no_dev/norm
trim_frame_size: 2048
trim_hop_size: 512
trim_silence: false
trim_threshold_in_db: 60
verbose: 1
version: 0.1.0
vocoder: null
win_length: null
window: hann
feat_list:
    mel: {}
trg_feat: "mel"
collater_type: "ARVCCollater"
trainer_type: "ARVCTrainer"
criterions:
  Seq2SeqLoss:
    bce_pos_weight: 10.0