diff options
author | Jean-Marc Valin <jmvalin@jmvalin.ca> | 2017-07-12 23:55:28 +0300 |
---|---|---|
committer | Jean-Marc Valin <jmvalin@jmvalin.ca> | 2017-10-06 00:40:27 +0300 |
commit | af93fbd55fd5c23a2492166816311d9f67df1b24 (patch) | |
tree | 7221fd8dd284dd593e4b3eb1a3ed9cee3b4fc926 /scripts | |
parent | f3cff05eeb83ec8c055b7331338d705af220358d (diff) |
Add RNN for VAD and speech/music classification
Based on two dense layers with a GRU layer in the middle
Diffstat (limited to 'scripts')
-rwxr-xr-x | scripts/dump_rnn.py | 57 | ||||
-rwxr-xr-x | scripts/rnn_train.py | 67 |
2 files changed, 124 insertions, 0 deletions
diff --git a/scripts/dump_rnn.py b/scripts/dump_rnn.py new file mode 100755 index 00000000..dd66403b --- /dev/null +++ b/scripts/dump_rnn.py @@ -0,0 +1,57 @@ +#!/usr/bin/python + +from __future__ import print_function + +from keras.models import Sequential +from keras.layers import Dense +from keras.layers import LSTM +from keras.layers import GRU +from keras.models import load_model +from keras import backend as K + +import numpy as np + +def printVector(f, vector, name): + v = np.reshape(vector, (-1)); + #print('static const float ', name, '[', len(v), '] = \n', file=f) + f.write('static const opus_int16 {}[{}] = {{\n '.format(name, len(v))) + for i in range(0, len(v)): + f.write('{}'.format(int(round(8192*v[i])))) + if (i!=len(v)-1): + f.write(',') + else: + break; + if (i%8==7): + f.write("\n ") + else: + f.write(" ") + #print(v, file=f) + f.write('\n};\n\n') + return; + +def binary_crossentrop2(y_true, y_pred): + return K.mean(2*K.abs(y_true-0.5) * K.binary_crossentropy(y_pred, y_true), axis=-1) + + +model = load_model("weights.hdf5", custom_objects={'binary_crossentrop2': binary_crossentrop2}) + +weights = model.get_weights() + +f = open('rnn_weights.c', 'w') + +f.write('/*This file is automatically generated from a Keras model*/\n\n') +f.write('#ifdef HAVE_CONFIG_H\n#include "config.h"\n#endif\n\n#include "mlp.h"\n\n') + +printVector(f, weights[0], 'layer0_weights') +printVector(f, weights[1], 'layer0_bias') +printVector(f, weights[2], 'layer1_weights') +printVector(f, weights[3], 'layer1_recur_weights') +printVector(f, weights[4], 'layer1_bias') +printVector(f, weights[5], 'layer2_weights') +printVector(f, weights[6], 'layer2_bias') + +f.write('const DenseLayer layer0 = {\n layer0_bias,\n layer0_weights,\n 25, 16, 0\n};\n\n') +f.write('const GRULayer layer1 = {\n layer1_bias,\n layer1_weights,\n layer1_recur_weights,\n 16, 12\n};\n\n') +f.write('const DenseLayer layer2 = {\n layer2_bias,\n layer2_weights,\n 12, 2, 1\n};\n\n') + +f.close() diff --git a/scripts/rnn_train.py b/scripts/rnn_train.py new file mode 100755 index 00000000..ffdaa1e7 --- /dev/null +++ b/scripts/rnn_train.py @@ -0,0 +1,67 @@ +#!/usr/bin/python + +from __future__ import print_function + +from keras.models import Sequential +from keras.models import Model +from keras.layers import Input +from keras.layers import Dense +from keras.layers import LSTM +from keras.layers import GRU +from keras.layers import SimpleRNN +from keras.layers import Dropout +from keras import losses +import h5py + +from keras import backend as K +import numpy as np + +def binary_crossentrop2(y_true, y_pred): + return K.mean(2*K.abs(y_true-0.5) * K.binary_crossentropy(y_pred, y_true), axis=-1) + +print('Build model...') +#model = Sequential() +#model.add(Dense(16, activation='tanh', input_shape=(None, 25))) +#model.add(GRU(12, dropout=0.0, recurrent_dropout=0.0, activation='tanh', recurrent_activation='sigmoid', return_sequences=True)) +#model.add(Dense(2, activation='sigmoid')) + +main_input = Input(shape=(None, 25), name='main_input') +x = Dense(16, activation='tanh')(main_input) +x = GRU(12, dropout=0.1, recurrent_dropout=0.1, activation='tanh', recurrent_activation='sigmoid', return_sequences=True)(x) +x = Dense(2, activation='sigmoid')(x) +model = Model(inputs=main_input, outputs=x) + +batch_size = 64 + +print('Loading data...') +with h5py.File('features.h5', 'r') as hf: + all_data = hf['features'][:] +print('done.') + +window_size = 1500 + +nb_sequences = len(all_data)/window_size +print(nb_sequences, ' sequences') +x_train = all_data[:nb_sequences*window_size, :-2] +x_train = np.reshape(x_train, (nb_sequences, window_size, 25)) + +y_train = np.copy(all_data[:nb_sequences*window_size, -2:]) +y_train = np.reshape(y_train, (nb_sequences, window_size, 2)) + +all_data = 0; +x_train = x_train.astype('float32') +y_train = y_train.astype('float32') + +print(len(x_train), 'train sequences. x shape =', x_train.shape, 'y shape = ', y_train.shape) + +# try using different optimizers and different optimizer configs +model.compile(loss=binary_crossentrop2, + optimizer='adam', + metrics=['binary_accuracy']) + +print('Train...') +model.fit(x_train, y_train, + batch_size=batch_size, + epochs=200, + validation_data=(x_train, y_train)) +model.save("newweights.hdf5") |