Artificial intelligence taught to speak in the voice of Bill Gates

Facebook has developed a synthesizer based on artificial intelligence that can imitate human intonation and even copy the voices of real people. In particular, the development team taught the assistant to speak with the voice of Microsoft's founder, Bill Gates.


Artificial intelligence taught to speak in the voice of Bill Gates
Artificial intelligence taught to speak in the voice of Bill Gates


What is known about new development? According to MIT Technology Review, Sean Vazquez and Mike Lewis from Facebook AI Research were able to overcome the limitations of systems using the machine learning system called MelNet.

Development is able not only to reproduce human intonations but also to imitate the voices of really existing people. In particular, the researchers "fake" the language of Bill Gates. Recordings of other voices with various intonations can be listened to via the link.

How does AI work? A new approach to Vasquez and Lewis is that they used spectrographs to study the system. This is a graphical record of the sound wave, which includes all the features of the language. In this form, it is much easier to teach artificial intelligence.