Amazon's neural TTS can model speaking styles with only a few hours of recordings

Scientists at Amazon have developed a novel method of modeling different speaking styles using AI. The results are pretty impressive.