Learning De-identified Representations of Prosody from Raw Audio

             Jack Weston 1 Raphal Lenain 1 Udeepa Meepegama 1 Emil Fristed 1

             Abstract               The phonetic problem has automatic speech recognition
                                (ASR) as its obvious use-case. In recent years, speech
    We propose a method for learning de-identified      representation learning has been increasingly dominated
    prosody representations from raw audio using       by self-sup ...