Project link: https://github.com/mozilla/DeepSpeech

Deep Speech is an open source speech-to-text engine, based on a research paper from Baidu. Speech-to-text is a very interesting problem and Deep Speech is one of the state-of-the-art methods for solving it. With the TensorFlow implementation of Mozilla, you can even learn how to use TensorFlow across more than one machine. However, there is still a problem that personal researchers can't access the same large scale speech-to-text datasets as a large company. So, even though we can use Deep Speech or implement it ourselves, it is still hard to have a good model for production.