-
Notifications
You must be signed in to change notification settings - Fork 0
/
wavProcessing.py
274 lines (242 loc) · 10.6 KB
/
wavProcessing.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
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
# This file is no longer needed for our project, but it might be useful to keep
# TODO: Consider deleting this file
from pyAudioAnalysis import audioFeatureExtraction
import matplotlib.pyplot as plt
import os
import numpy
from pydub import AudioSegment
def readAudioFile(path):
"""Reads an audio file located at specified path and returns a numpy array of audio samples
NOTE: This entire function was ripped from pyAudioAnalysis.audioBasicIO.py
All credits to the original author, Theodoros Giannakopoulos.
The existing audioBasicIO.py relies on broken dependencies, so it is much more reliable to rip the only function
we need to process WAV files
Paramters
----------
path : str
The path to a given audio file
Returns
----------
Fs : int
Sample rate of audio file
x : numpy array
Data points of the audio file"""
extension = os.path.splitext(path)[1]
try:
# Commented below, as we don't need this
# #if extension.lower() == '.wav':
# #[Fs, x] = wavfile.read(path)
# if extension.lower() == '.aif' or extension.lower() == '.aiff':
# s = aifc.open(path, 'r')
# nframes = s.getnframes()
# strsig = s.readframes(nframes)
# x = numpy.fromstring(strsig, numpy.short).byteswap()
# Fs = s.getframerate()
if extension.lower() == '.mp3' or extension.lower() == '.wav' or extension.lower() == '.au' or extension.lower() == '.ogg':
try:
audiofile = AudioSegment.from_file(path)
except:
print("Error: file not found or other I/O error. "
"(DECODING FAILED)")
return -1 ,-1
if audiofile.sample_width == 2:
data = numpy.fromstring(audiofile._data, numpy.int16)
elif audiofile.sample_width == 4:
data = numpy.fromstring(audiofile._data, numpy.int32)
else:
return -1, -1
Fs = audiofile.frame_rate
x = numpy.array(data[0::audiofile.channels]).T
else:
print("Error in readAudioFile(): Unknown file type!")
return -1, -1
except IOError:
print("Error: file not found or other I/O error.")
return -1, -1
if x.ndim == 2:
if x.shape[1] == 2:
x = x.flatten()
return Fs, x
def get_sig(filename):
"""Gets a signal from an audio file
Parameters
----------
filename : string
Name of the WAV audio file, including extension
Returns
----------
rate : int
Sample rate of the audio file
signal : numpy array
NumPy array containing all sample points in the audio file.
The NumPy dtype in the array depends on the format of the WAV file."""
(rate, data) = readAudioFile(filename)
return rate, data
def get_st_features(signal, rate, window_step=0.025, window_length=0.05):
"""Computes all 34 features for each window in a given signal
Parameters
----------
signal : numpy array
All sample points for the audio signal
Can be any type of number
rate : int
Sample rate of the audio signal, in Hz
window_step : float
Time step between each successive window, in seconds
Default: 0.025 (25 ms)
window_length : float
Length of each window, in seconds
Should generally be greater than windowStep to allow for overlap between frames
Default: 0.05 (50 ms)
Returns
----------
features : numpy array
NumPy array of size (number of windows) * 34
Each row in mfcc_features contains all the features for a single frame
feature_names : [str]
Names of each feature located at specified index"""
sample_step = int(rate*window_step)
sample_length = int(rate*window_length)
(features, feature_names) = audioFeatureExtraction.stFeatureExtraction(signal, rate, sample_length, sample_step)
return features, feature_names
def relevant_indexes(data, min_threshold):
"""Finds first and last index where data > min_threshold
To find the start and end indexes of the frames where there is some noise
Could be useful to take many audio clips and find the lowest start index and highest end index common between
all audio clips. This would be useful if the ML code must take a fixed # of input layer data points
Parameters
----------
data : numpy array
Energy levels of multiple frames
min_threshold : float
Minimum threshold value that each data is compared to
Returns
----------
start_index : int
First index in data with a value greater than min_threshold
end_index : int
Last index in data with a value greater than min_threshold"""
start_index = 1
end_index = len(data) - 1
for i in range(len(data)):
if data[i] > min_threshold:
start_index = i
break
for i in range(len(data)):
if data[::-1][i] > min_threshold:
end_index = i
break
return start_index, end_index
def make_line_plot(data, x_label="Data", y_label="Data Point"):
"""Creates a line plot of data, where each point on the plot is (i, data[i])
Parameters
----------
data : numpy array
Any type of homogeneous numerical data
x_label : str
The label to put on the independent axis
y_label : str
The label to put on the dependent axis
Returns
----------
None"""
y = data
x = range(len(y))
plt.xlabel(x_label)
plt.ylabel(y_label)
plt.plot(x, y)
plt.show()
def get_trimmed_features(words, num_recordings, base_path="", energy_threshold=0.001):
"""Calculates features for a list of words, returning trimmed data based on a frame energy threshold
Assumes all audio recordings are in the same directory base_path, and all recordings are WAV format.
Calculates features for every recording and returns them in a hierarchical array to be fed into a neural network.
The number of frames for each word type is the same for all recordings of that word type, as determined by
the energy threshold for each frame.
Parameters
----------
words : [str]
A list of distinct words
It is assumed that audio files will have path base_path/(word)(num).wav
Where word is one of the words in the words parameter
num_recordings : [int]
A list of integers >= 1
List must have same length as words
For word words[i], there should be num_recordings[i] distinct recordings/files of that word
It is assumed that audio files will have path base_path/(word)(num).wav
Where num is in the range of 1 to num_recordings
base_path : str
The base path that will be appended to all audio file paths as a prefix
This is where the directory of audio files would be specified
energy_threshold : float
Minimum energy for a given frame to be considered relevant
i.e. if a frame is loud enough or contains enough information to impact the data set
Returns
----------
features_by_word : numpy array
Cell array of same length as words
Ordering of cells is determined by the order of the words in words parameter
The ith cell has num_recordings[i] elements
Each element in a cell is an array of equal lengths, with each element in said array containing all relevant
frames
Within each frame are the 34 features extracted by pyAudioAnalysis"""
features_by_word = []
for i in range(len(words)):
indexes = []
feature_array = []
for j in range(1, num_recordings[i] + 1):
# Determine the path
path = base_path + words[i] + str(j) + ".wav"
(rate, data) = get_sig(path)
# features is all the audio features for a given file
features = get_st_features(data, rate)[0]
# features[1] is total frame energies
# energy threshold of 0.001 is arbitrary
indexes.append(relevant_indexes(features[1], energy_threshold))
# Add features for this specific audio file to the feature array for this word
feature_array.append(features)
# Finds the minimum index of all start indexes
min_index = sorted(indexes, key=lambda x: x[0])[0][0]
# Finds the max index of all end indexes
max_index = sorted(indexes, key=lambda x: x[1])[::-1][0][1]
# Debug print statements commented out
# print("min, max index for word", words[i])
# print(min_index, max_index)
# Only take the frames between min index and max index for each sample word
# Note: Potential for a bug; if maxIndex is outside the length of its frame array
# To fix, need to pad the shorter recordings with extra data
features_by_word.append([x[0:34, min_index:max_index].transpose() for x in feature_array])
# print(numpy.shape(features_by_word[i]))
# features_by_word is an array of len(words) cells
# Each cell has num_recordings[i] elements corresponding to the number of recordings of each word words[i]
# Each recording has the same number of frames for a given word, as determined by minIndex and maxIndex
# for a given word.
# Finally, each frame contains the 34 features from that frame's raw data samples
return features_by_word
# word_list = ["light", "off", "on", "slack", "tv"]
# # Could change this to numbers between 1 and 30 to see how it handles more or less data
# nums = [30, 30, 30, 30, 30]
# # The base_directory might be different for windows users
# base_directory = "ModernOTData/"
# output = get_trimmed_features(word_list, nums, base_directory)
#
# # energy_values is a sequential list of all energy values over all recordings
# energy_values = []
#
# # Should print 5
# print("There are", len(output), "different words")
# for word_num in range(len(output)):
# # Should print 30
# print("There are", len(output[word_num]), "different recordings for word", word_list[word_num])
# for recording_num in range(len(output[word_num])):
# # Print number of frames for each recording
# # Should be equal for all words
# print("# frames:", len(output[word_num][recording_num]), "in recording #", str(recording_num+1), "for word",
# word_list[word_num])
# for frame in output[word_num][recording_num]:
# # Should be 34 features for each frame
# # print(len(frame))
# # frame[1] is the energy for that frame
# energy_values.append(frame[1])
#
# # Sample plot of energies across every recording
# make_line_plot(energy_values, "Frame Number", "Energy")