Speech auto tagging

identify the language's attribute(such as language type,gender,accent etc.) use keras. now only implementation language type classify.

1. data prepare

use librispeech and THCHS30 as english and chinese speech corpus

run in shell: python data_prepare

2. train

run: python train.py

3. test

run : python test.py