Underdog shows off a 32MB speech model in new demo

Despite its small size, Underdog's new model can generate both speech and music.

Summary aggregated from TechCrunch's public RSS feed. The full reporting belongs to TechCrunch — please read it on their site.