egs/egs_bases/tts/dataset_params.yaml

audio_num_mel_bins: 80
audio_sample_rate: 22050
hop_size: 256  # For 22050Hz, 275 ~= 12.5 ms (0.0125 * sample_rate)
win_size: 1024  # For 22050Hz, 1100 ~= 50 ms (If None, win_size: fft_size) (0.05 * sample_rate)
fft_size: 1024  # Extra window size is filled with 0 paddings to match this parameter
fmin: 80  # Set this to 55 if your speaker is male! if female, 95 should help taking off noise. (To test depending on dataset. Pitch info: male~[65, 260], female~[100, 525])
fmax: 7600  # To be increased/reduced depending on data.
f0_min: 80
f0_max: 800
griffin_lim_iters: 30
pitch_extractor: parselmouth
num_spk: 1
mel_vmin: -6
mel_vmax: 1.5
loud_norm: false

raw_data_dir: ''
processed_data_dir: ''
binary_data_dir: ''
preprocess_cls: ''
binarizer_cls: data_gen.tts.base_binarizer.BaseBinarizer
preprocess_args:
  nsample_per_mfa_group: 1000
  # text process
  txt_processor: en
  use_mfa: true
  with_phsep: true
  reset_phone_dict: true
  reset_word_dict: true
  add_eos_bos: true
  # mfa
  mfa_group_shuffle: false
  mfa_offset: 0.02
  # wav processors
  wav_processors: [ ]
  save_sil_mask: true
  vad_max_silence_length: 12
binarization_args:
  shuffle: false
  with_wav: false
  with_align: true
  with_spk_embed: false
  with_f0: true
  with_f0cwt: false
  with_linear: false
  trim_eos_bos: false
  min_sil_duration: 0.1
  train_range: [ 200, -1 ]
  test_range: [ 0, 100 ]
  valid_range: [ 100, 200 ]
word_dict_size: 10000
pitch_key: pitch