Voice assistant Google Assistant will sound more realistic thanks to WaveNet technology
The company DeepMind, which in 2014 is a subsidiary of Google, has created a new technology of synthesis of human speech WaveNet, using his experience with artificial intelligence.
Now WaveNet technology to be used to generate speech in the voice assistant Google Assistant, which will sound much more realistic. Lost native accent, Google Assistant will speak as a living person.
WaveNet is different from other technologies for the synthesis of the human voice, this technology can generate sounds, turning up to 16,000 more suitable samples per second. Within 12 months, the developers were trained WaveNet find a more correct intonation, now the system understands what the correct tone to apply in a particular case, so it sounded realistic.
At the initial stage WaveNet will be used for the English and Japanese versions of Google Assistant in the future, the list of languages will be expanded.