Wakenet

no secret you'll feel better if you cry When wakened from a bad dream Don't you sometimes think

Unhoped-for entrance thro' some friend within, One clear idea, wakened in the breast By memory's magic, lets in

Uno de los camareros de la cafetería de la biblioteca de la UPV-EHU de Leioa nos cuenta si tienen productos para personas intolerantes

loves you best? When the man with the answer Has wakened you, and warned you, and called you to the

This was generated by a Wavenet trained on the same representation and dataset as PerformanceRNN. It was primed with the first 20 seconds of Sonata in F Major, K280 by Mozart (not included in the track). This model is relatively small (receptive field size of 1024), and generally has worse output than PerformanceRNN. It's a proof of concept showing that Wavenet works on this task. I plan to follow up with a deeper wavenet in the future. All the code is here (although I trained the network locally): https://colab.research.google.com/drive/1-MdCFoz7IV9q4HhVt5DWTHOoYhyk2Hqu?usp=sharing

Piano music audio generated using a Wavenet model implementation through highest model prediction probability.

See github.com/basveeling/wavenet Sample Epoch - 00037 04s Sample - Temp - 0.001 Seed - 1215123

lying kind (Ft. kuru)

not sorry