NeuralSeq/egs/datasets/audio/emotion/base_text2mel.yaml

raw_data_dir: 'data/raw/ESD'
processed_data_dir: 'data/processed/emotion'
binary_data_dir: 'data/binary/emotion'
pre_align_cls: egs.datasets.audio.emotion.pre_align.EmoPreAlign
audio_sample_rate: 16000
binarization_args:
  shuffle: true
binarizer_cls: data_gen.tts.base_binarizer_emotion.EmotionBinarizer
use_spk_id: true
test_num: 200
num_spk: 10
pitch_type: frame
min_frames: 128
num_test_samples: 30
mel_loss: "ssim:0.5|l1:0.5"
vocoder_ckpt: ''
use_emotion: true