NeuralSeq/egs/egs_bases/tts/vocoder/base.yaml

base_config: ../base.yaml
binarization_args:
  with_wav: true
  with_spk_embed: false
  with_align: false
  with_word: false
  with_txt: false

###########
# train and eval
###########
max_samples: 25600
max_sentences: 5
max_valid_sentences: 1
max_updates: 1000000
val_check_interval: 2000

###########################################################
#                FEATURE EXTRACTION SETTING               #
###########################################################
fft_size: 1024           # FFT size.
hop_size: 256            # Hop size.
win_length: null         # Window length.
# If set to null, it will be the same as fft_size.
window: "hann"           # Window function.
num_mels: 80             # Number of mel basis.
fmin: 80                 # Minimum freq in mel basis calculation.
fmax: 7600               # Maximum frequency in mel basis calculation.
aux_context_window: 0 # Context window size for auxiliary feature.
use_pitch_embed: false

generator_grad_norm: 10    # Generator's gradient norm.
discriminator_grad_norm: 1 # Discriminator's gradient norm.
disc_start_steps: 40000 # Number of steps to start to train discriminator.