this repository contains end-to-end pipe line to train different speech data provided by google, evaluate testing data, and submite to kaggle competition. (https://www.kaggle.com/c/tensorflow-speech-recognition-challenge)
clone repo. :- git clone https://github.com/parthvadhadiya/TensorFlow-Speech-Recognition-Challenge
download dataset :- https://www.kaggle.com/c/tensorflow-speech-recognition-challenge/data
=> Python3 -3.5.2
=> NumPy -1.13.1
for store and manupulate multidimensional array,
installing numpy:- sudo pip3 install numpy (obviously for windows you can enter without sudo command).
=> librosa -0.5.1
for audio analysis
installing librosa:- sudo pip3 install librosa(you can also use scipy's wav module)
=> keras -2.0.6
high-level neural networks API
installing keras:- https://keras.io/#installation
=> tqdm -4.19.4
an awsome progress meter in terminal
intalling tqdm :- sudo pip3 install tqdm