I've been working on this for the past few weeks and was surprised to see pretty polished competitor posted yesterday: https://news.ycombinator.com/item?id=19454863

Oh well, great minds think alike!

EDIT: Similarly, this uses Google's Text to Speech (https://cloud.google.com/text-to-speech/). If this gets some traction I would love to develop my own TTS model using WaveNet (https://deepmind.com/blog/wavenet-generative-model-raw-audio...), since the TTS API is super expensive.