Speech to text things
Baidu’s Deep Speech is very popular Many repos out there implemented the model in tensorflow. There are many datasets can use and train. Mozilla has 12GB of data Kur looks fancy but doesn’t seem to be necessary. It uses yaml to define the models, with help of Jinja templating. Too many “Kur is super easy and flexible and powerful” promises. I don’t feel they really know machine learning.
Kur, Keras, and Tensorflow
Code is available on GitHub A simple overview of the same model written with three machine learning frameworks Kur, Keras, and Tensorflow. An exploration of a data pipeline for Tensorflow using TFRecords. A year or so ago when Tensorflow came out I, like many others, downloaded it, and tried to…