Using deep neural networks, Wavenet was given raw input of enormous speeches as a dataset. By using training data, it could generate audio very similar to real human speech. Though accuracy was not 100% but yet t was considered better than Google’s Text-to-Speech (TTS). Read more

September, 2016