Episode 21: Look Who's Talking

What's next in text-to-speech?

In December 2017, Google presented a paper that showed a new technological advance in text-to-speech. Until now, Google says, pre-recorded words stitched together doesn’t do the trick. The computerized voice comes off sounding fake. No kidding. The new approach that Google took resulted in Tacotron 2 - a new method of parsing text to speech that can replicate many of the nuances of human speech.

Samples of Google's Tacitron 2: https://google.github.io/tacotron/publications/tacotron2/index.html

Online text-to-speech resources used in this episode:
Click here to listen on Stitcher: http://www.stitcher.com/s?fid=152797&refid=stpr
Click here to listen on iTunes: https://itunes.apple.com/il/podcast/y-karp-why-not/id1291793305
Click here to listen on TuneIn: https://tunein.com/radio/Y-Karp-Why-Not-p1038454/

Or play the episode directly from here:



Yossi's Magazine: Everyone Needs One of These www.ykarp.com/magazine
Follow on Twitter: @ykarp
Follow on Google+: +Yossi Karp
Subscribe to Y. Karp? Why Not! or follow on Facebook
Add this blog to your RSS feed reader

Comments

Popular posts from this blog

Act Your Age, Not Your Shoe Size

Playing With My Mind

Lessons from a Couch