1 paper
Sewade Ogun, Vincent Colotte, Emmanuel Vincent
Flow-based generative models are widely used in text-to-speech (TTS) systems to learn the distribution of audio features (e.g., Mel-spectrograms) given the input tokens and to samp…