Really compelling work here! The audio samples include a “human or synthesis” challenge, which… is challenging. google.github.io/tacotron/publi
Quote Tweet
New work on text-to-speech from @GoogleResearch, done jointly by members of the Google Brain and Machine Perception research teams. From the Arxiv abstract:
"Our model achieves a mean opinion score of 4.53 comparable to a MOS of 4.58 for professionally recorded speech." twitter.com/googleresearch
Show this thread
