May 01, 2021
Learning de-identified representations of prosody from raw audio has been accepted at ICML 2021.