Python Deep Learning Cookbook

How to do it...

1. We start by importing keras and other libraries, as follows:

import globimport numpy as npimport randomimport librosafrom sklearn.model_selection import train_test_splitimport kerasfrom keras.layers import LSTM, Dense, Dropout, Flattenfrom keras.models import Sequentialfrom keras.optimizers import Adamfrom keras.callbacks import EarlyStopping, ModelCheckpoint

2. First, we set the location of the training files, and for reproducibility, we also set a SEED:

SEED = 2017DATA_DIR = 'Data/spoken_numbers_pcm/' 

3. The training files are located in a directory, and we will be using a batch generator to retrieve them. First, we split the training files into a training and a validation set:

files = glob.glob(DATA_DIR + "*.wav")X_train, ...