Google’s WaveNet has been recently improved.
(Mirror Daily, United States) Google is one of the big names in the technological domain. As everything related to artificial technology is rapidly advancing, the great company couldn’t possibly stay behind. The latest news from Google is that its system, WaveNet, now has an improved voice program, which makes it sound more human-like than before, and than any other similar robots.
WaveNet is a system which belongs to a British company, called DeepMind. Google owns the company since 2014.
Engineers worked on the improvement of the system. The process behind it is based on concatenative speech. This means that the system works using real human speech. Several phrases and utterances have been recorded and put together for the system to function correctly. It uses waveforms to function (waveforms are the audible forms of words and phrases).
However, there is more about Google’s WaveNet. It doesn’t only work based on the collection of recordings it possesses, but it is also able to produce utterances of its own. Another interesting ability is that it can also use more than one voice.
Engineers thought about an additional feature that they could endow WaveNet with. It seems like the system can analyze and reproduce pieces of instrumental music. For the time being, it works well with piano instrumentals, but the developers intend to add other musical instruments as well. They also consider the possibility of WaveNet being able to create its own melodies.
As far as the systems’ interaction with humans is concerned, it has also been tested. English and Mandarin Chinese speakers took part at the trials set by the developers. The participants declared themselves satisfied by the accuracy of the voice, and by the general interaction with the artificial form of intelligence from Google.
It seems like both developers, and potential users are pleased with the improved technology by DeepMind and Google. However, the creators inform the public that the technology is not to be released for the moment. In other words, it is only a prototype. The developers consider that the human-like voice is only one checked improvement in a long list that remains to be analyzed.
WaveNet has already proved itself able to fill the gap between robotic intelligence with weird sounding voice and natural human speech. Perhaps the future will bring this improved system on a wide variety of devices.
Image courtesy of: Wikipedia