Our current deep learning approach leverages more data, more computational power, and novel techniques to deliver our most advanced speech synthesis model