3 ms·
Since this is around 18 months old, does anyone know what's the current state of the art?
by tziki 9y ago
Since this is around 18 months old, does anyone know what's the current state of the art?
- yomansat 9y agoI found this from ~3 months ago, the recent addition to github: https://google.github.io/tacotron/publications/uncovering_latent_style_factors_for_expressive_speech_synthesis/index.html https://google.github.io/tacotron/publications/uncovering_la... Seems to do with manipulating the speed styles in how we speak. Listen to the "neutral" blue lagoon, vs token 6 (tired sound), vs mix w/ token 7 (prominence)" (very confident). Paper: https://arxiv.org/pdf/1711.00520.pdf https://arxiv.org/pdf/1711.00520.pdf