Learned Representations of Tone Contours in GANs
Generative adversarial networks (GANs) have been applied to many speech generation tasks. This project trains GAN model on Mandarin Chinese syllables and determines whether the model can represent each of the four tonal contours in Mandarin are represented in its latent space.
Message To SponsorThank you so much for supporting my summer research.
Mentor: Gasper Begus
Sponsor: Holmes-Olsen Fund