Doing research to see where we currently are with faking voice audio with neural networks/deep learning. Learning to create voices from YouTube clips, and trying to see how quickly we can do new voices. In this case, I've used a Deep Convolutional Text to Speech (DCTTS) model to produce pretty darn good results. The voices in the first 2 minutes are all fake.
Jul-12-2019, 16:46:15 GMT