Google releases text-to-speech technology "Cloud Text-to-Speech" created by DeepMind, making it available to anyone

byJacek Dylag

The Google Text - to - Speech engine used for Google Assistant and Google Map navigation has been released for developers. This Text-to-Speech engine is a voice input / output technology created by Deep Mind "WaveNetIncluding high fidelity sound using ", you can use various methods such as" correspondence of call center "and" utilization in IoT device ".

Google Cloud Platform Blog: Introducing Cloud Text-to-Speech powered by DeepMind WaveNet technology
https://cloudplatform.googleblog.com/2018/03/introducing-Cloud-Text-to-Speech-powered-by-Deepmind-WaveNet-technology.html

Google's "Cloud Text-to-Speech" is a beta version at the time of article creation, but it can be accessed from the following link.

Cloud Text-to-Speech - Speech Synthesis | Google Cloud
https://cloud.google.com/text-to-speech/


Cloud Text-to-Speech supports 12 languages ​​including English and 32 kinds of voice. Developers can adjust pitch, vocalization speed, volume gain of MP3 or WAV.

WaveNetTechnology released in 2016In the first model, it took 1 second to make a waveform of 0.02 seconds, but as of March 2018, it became 1000 times faster than the original model, and a voice of 1 second length was reproduced in 50 milliseconds It is possible to generate.


It is also a point that not only speed but also high sound quality.Average opinion scoreAs a result, the sound of WaveNet is better than the standard score of 20% or better, and it is shown that it is approaching the actual human voice.