lstm_speaker_verification

1. this project is an implement of a methode proposed by Google in ICASSP 2018

"ATTENTION-BASED MODELS FOR TEXT-DEPENDENT SPEAKER VERIFICATION"

pdf of the paper:https://arxiv.org/pdf/1710.10470.pdf

2. execute this progrmme by using the command as follow:

python train.py

3. the file input_data.py is used for exitracting mfcc features and prepare the input data of training LSTM

the file model.py is used for construction of neural network