egs/datasets/audio/libritts/base_text2mel.yaml

raw_data_dir: 'data/raw/LibriTTS'
processed_data_dir: 'data/processed/libritts'
binary_data_dir: 'data/binary/libritts'
pre_align_cls: egs.datasets.audio.libritts.pre_align.LibrittsPreAlign
binarization_args:
  shuffle: true
use_spk_id: true
test_num: 200
num_spk: 2320
pitch_type: frame
min_frames: 128
num_test_samples: 30
mel_loss: "ssim:0.5|l1:0.5"
vocoder_ckpt: ''