TADA: Fast, Reliable Speech Generation Through Text-Acoustic SynchronizationComments on Hacker News | Source