-
Notifications
You must be signed in to change notification settings - Fork 6
/
bare_bones_asr.py
185 lines (161 loc) · 6.69 KB
/
bare_bones_asr.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
import tensorflow as tf
import numpy as np
import librosa
from string import ascii_lowercase
class ASR(tf.keras.Model):
'''
Class for defining the end to end ASR model.
This model consists of a 1D convolutional layer followed by a bidirectional LSTM
followed by a fully connected layer applied at each timestep.
This is a bare-bones architecture.
Experiment with your own architectures to get a good WER
'''
def __init__(self, filters, kernel_size, conv_stride, conv_border, n_lstm_units, n_dense_units):
super(ASR, self).__init__()
self.conv_layer = tf.keras.layers.Conv1D(filters,
kernel_size,
strides=conv_stride,
padding=conv_border,
activation='relu')
self.lstm_layer = tf.keras.layers.LSTM(n_lstm_units,
return_sequences=True,
activation='tanh')
self.lstm_layer_back = tf.keras.layers.LSTM(n_lstm_units,
return_sequences=True,
go_backwards=True,
activation='tanh')
self.blstm_layer = tf.keras.layers.Bidirectional(self.lstm_layer, backward_layer=self.lstm_layer_back)
self.dense_layer = tf.keras.layers.Dense(n_dense_units)
def call(self, x):
x = self.conv_layer(x)
x = self.blstm_layer(x)
x = self.dense_layer(x)
return x
def compute_ctc_loss(logits, labels, logit_length, label_length):
'''
function to compute CTC loss.
Note: tf.nn.ctc_loss applies log softmax to its input automatically
:param logits: Logits from the output dense layer
:param labels: Labels converted to array of indices
:param logit_length: Array containing length of each input in the batch
:param label_length: Array containing length of each label in the batch
:return: array of ctc loss for each element in batch
'''
return tf.nn.ctc_loss(
labels=labels,
logits=logits,
label_length=label_length,
logit_length=logit_length,
logits_time_major=False,
unique=None,
blank_index=-1,
name=None
)
def create_spectrogram(signals):
'''
function to create spectrogram from signals loaded from an audio file
:param signals:
:return:
'''
stfts = tf.signal.stft(signals, frame_length=200, frame_step=80, fft_length=256)
spectrograms = tf.math.pow(tf.abs(stfts), 0.5)
return spectrograms
def generate_input_from_audio_file(path_to_audio_file, resample_to=8000):
'''
function to create input for our neural network from an audio file.
The function loads the audio file using librosa, resamples it, and creates spectrogram form it
:param path_to_audio_file: path to the audio file
:param resample_to:
:return: spectrogram corresponding to the input file
'''
# load the signals and resample them
signal, sample_rate = librosa.core.load(path_to_audio_file)
if signal.shape[0] == 2:
signal = np.mean(signal, axis=0)
signal_resampled = librosa.core.resample(signal, sample_rate, resample_to)
# create spectrogram
X = create_spectrogram(signal_resampled)
# normalisation
means = tf.math.reduce_mean(X, 1, keepdims=True)
stddevs = tf.math.reduce_std(X, 1, keepdims=True)
X = tf.divide(tf.subtract(X, means), stddevs)
return X
def generate_target_output_from_text(target_text):
'''
Target output is an array of indices for each character in your string.
The indices comes from a mapping that will
be used while decoding the ctc output.
:param target_text: (str) target string
:return: array of indices for each character in the string
'''
space_token = ' '
end_token = '>'
blank_token = '%'
alphabet = list(ascii_lowercase) + [space_token, end_token, blank_token]
char_to_index = {}
for idx, char in enumerate(alphabet):
char_to_index[char] = idx
y = []
for char in target_text:
y.append(char_to_index[char])
return y
def train_sample(x, y, optimizer, model):
'''
function perform forward and backpropagation on one batch
:param x: one batch of input
:param y: one batch of target
:param optimizer: optimizer
:param model: object of the ASR class
:return: loss from this step
'''
with tf.GradientTape() as tape:
logits = model(x)
labels = y
logits_length = [logits.shape[1]]*logits.shape[0]
labels_length = [labels.shape[1]]*labels.shape[0]
loss = compute_ctc_loss(logits, labels, logit_length=logits_length, label_length=labels_length)
loss = tf.reduce_mean(loss)
grads = tape.gradient(loss, model.trainable_variables)
optimizer.apply_gradients(zip(grads, model.trainable_variables))
return loss
def train(model, optimizer, X, Y, epochs):
'''
function to train the model for given number of epochs
Note:
For this example, I am passing a single batch of input to this function
Therefore, the loop for iterating through batches is missing
:param model: object of class ASR
:param optimizer: optimizer
:param X:
:param Y:
:param epochs:
:return: None
'''
for step in range(1, epochs):
loss = train_sample(X, Y, optimizer, model)
print('Epoch {}, Loss: {}'.format(step, loss))
if __name__ == '__main__':
sample_call = 'sample.wav'
transcript = 'MISTER QUILTER IS THE APOSTLE OF THE MIDDLE CLASSES AND WE ARE GLAD TO WELCOME HIS GOSPEL'.lower()
X = generate_input_from_audio_file(sample_call)
X = tf.expand_dims(X, axis=0) # converting input into a batch of size 1
y = generate_target_output_from_text(transcript)
y = tf.expand_dims(tf.convert_to_tensor(y), axis=0) # converting output to a batch of size 1
print('Input shape: {}'.format(X.shape))
print('Target shape: {}'.format(y.shape))
model = ASR(200, 11, 2, 'valid', 200, 29)
optimizer = tf.keras.optimizers.Adam()
train(model, optimizer, X, y, 100)
# getting the ctc output
ctc_output = model(X)
ctc_output = tf.nn.log_softmax(ctc_output)
# greedy decoding
space_token = ' '
end_token = '>'
blank_token = '%'
alphabet = list(ascii_lowercase) + [space_token, end_token, blank_token]
output_text = ''
for timestep in ctc_output[0]:
output_text += alphabet[tf.math.argmax(timestep)]
print(output_text)
print('\n\nNote: Applying a good decoder on this output will give you readable output')