Project link: https://github.com/ibab/tensorflow-wavenet
WaveNet is a neural network architecture for audio generation from Google Deep Mind. WaveNet is trained to generate raw audio waveform and has shown good results for text-to-speech and audio generation. According to Deep Mind, WaveNet reduced the gap between the previous methods and human-level performance by over 50% in text-to-speech problems for both US English and Mandarin Chinese.