text_to_speech/egs/egs_bases/tts/base.yaml

# task
base_config:
  - ../config_base.yaml
  - ./dataset_params.yaml

#############
# dataset in training
#############
endless_ds: true
min_frames: 0
max_frames: 1548
frames_multiple: 1
max_input_tokens: 1550
ds_workers: 0

#########
# model
#########
use_spk_id: false
use_spk_embed: false
mel_losses: "ssim:0.5|l1:0.5"

###########
# optimization
###########
lr: 0.0005
scheduler: warmup # rsqrt|warmup|none
warmup_updates: 4000
optimizer_adam_beta1: 0.9
optimizer_adam_beta2: 0.98
weight_decay: 0
clip_grad_norm: 1
clip_grad_value: 0


###########
# train and eval
###########
use_word_input: false
max_valid_sentences: 1
max_valid_tokens: 60000
valid_infer_interval: 10000
train_set_name: 'train'
train_sets: ''
valid_set_name: 'valid'
test_set_name: 'test'
num_valid_plots: 10
test_ids: [ ]
test_input_yaml: ''
vocoder: HifiGAN
vocoder_ckpt: ''
profile_infer: false
out_wav_norm: false
save_gt: true
save_f0: false
gen_dir_name: ''