-
Notifications
You must be signed in to change notification settings - Fork 2
/
SpatialiseTEST.py
184 lines (145 loc) · 5.84 KB
/
SpatialiseTEST.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
# -*- coding: utf-8 -*-
"""
Created on Sun Mar 21 18:43:20 2021
@author: Waradon Senzt Phokhinanan
"""
############################################################################################
import math
import librosa
import scipy.io
import matplotlib.pyplot as plt
from scipy import signal
import numpy as np
import os
import soundfile as sf
############################################################################################
def AdjustNoiseBySNR(signal,noise,SNR):
RMS_s = math.sqrt(np.mean(signal**2))
RMS_n = math.sqrt(RMS_s**2/(pow(10,SNR/10)))
RMS_nd = math.sqrt(np.mean(noise**2))
noise = noise*(RMS_n/RMS_nd)
return noise
############################################################################################
def NoiseTestingImport():
NoiseData = []
NoiseList = os.listdir('./NoiseTEST')
for NoiseX in NoiseList:
NoisePATH = './NoiseTEST/' + NoiseX
(NoiseI, rate) = librosa.load(NoisePATH, sr=16000)
NoiseI /= np.max(np.abs(NoiseI),axis=0)
NoiseData.append(NoiseI)
return NoiseData
############################################################################################
def spatialise37azimuths(speech_name, SNR, NoiseSig, NoiseNum):
Selected_speech = '.' + speech_name
(sig, rate) = librosa.load(Selected_speech, sr=16000)
sig /= np.max(np.abs(sig),axis=0)
MITmat = scipy.io.loadmat('hrir_MIT.mat')
HRIRMIT = MITmat['hrir_MIT']
HRIR_L0ele = HRIRMIT[:,:,0]
HRIR_R0ele = HRIRMIT[:,:,1]
HRIR_L0ele = librosa.resample(HRIR_L0ele, 44100, 16000)
HRIR_R0ele = librosa.resample(HRIR_R0ele, 44100, 16000)
HRIRLeft0 = HRIR_L0ele[19,:]
HRIRRight0 = HRIR_R0ele[19,:]
NewSigLEFT0 = signal.convolve(sig, HRIRLeft0)
NewSigRIGHT0 = signal.convolve(sig, HRIRRight0)
NoiseAdjustSNRLEFT=AdjustNoiseBySNR(NewSigLEFT0,NoiseSig,SNR)
NoiseAdjustSNRRIGHT=AdjustNoiseBySNR(NewSigRIGHT0,NoiseSig,SNR)
################ Spatialise 37 azimuths from -90 to 90 ################
ILDIPD_Feature = []
ILDIPD_Label = []
for x in range(0, 37):
HRIRLeftEx = HRIR_L0ele[x,:]
HRIRRightEx = HRIR_R0ele[x,:]
NewSigLEFT = signal.convolve(sig, HRIRLeftEx)
NewSigRIGHT = signal.convolve(sig, HRIRRightEx)
NewSigLEFT = NewSigLEFT+NoiseAdjustSNRLEFT[0:len(NewSigLEFT)]
NewSigRIGHT = NewSigRIGHT+NoiseAdjustSNRRIGHT[0:len(NewSigRIGHT)]
# #You can uncomment this section to listen the sptialised speech
# #test extraction
# NewSTERIO = np.array([NewSigLEFT,NewSigRIGHT])
# NewSTERIO = np.transpose(NewSTERIO)
# speech_nameC = speech_name.replace('/', '_')
# Exportname = 'Export-BinSP' + str(x) + '_' + str(NoiseNum) + '_' + str(speech_nameC)
# sf.write(Exportname, NewSTERIO, 16000, 'PCM_24')
################ Feature Extraction ################
STFTLeft = librosa.stft(NewSigLEFT, n_fft=640, hop_length=320, win_length=640, window='hamm', center=False, dtype=None, pad_mode='reflect')
STFTRight = librosa.stft(NewSigRIGHT, n_fft=640, hop_length=320, win_length=640, window='hamm', center=False, dtype=None, pad_mode='reflect')
IPD = np.angle(STFTRight/STFTLeft)
ILD = 20*np.log10(np.abs(STFTRight)/np.abs(STFTLeft))
IPDILD = np.array([IPD[:,20:70], ILD[:,20:70]])
IPDILD = np.moveaxis(IPDILD, 0, -1)
ILDIPD_Feature.append(IPDILD)
ILDIPD_Label.append(x)
ILDIPD_Feature = np.array(ILDIPD_Feature)
ILDIPD_Label = np.array(ILDIPD_Label)
return ILDIPD_Feature, ILDIPD_Label
############################################################################################
# MAIN PROGRAMME ###########################################################################
############################################################################################
azimuthdict = {
0: "-90",
1: "-85",
2: "-80",
3: "-75",
4: "-70",
5: "-65",
6: "-60",
7: "-55",
8: "-50",
9: "-45",
10: "-40",
11: "-35",
12: "-30",
13: "-25",
14: "-20",
15: "-15",
16: "-10",
17: "-5",
18: "0",
19: "5",
20: "10",
21: "15",
22: "20",
23: "25",
24: "30",
25: "35",
26: "40",
27: "45",
28: "50",
29: "55",
30: "60",
31: "65",
32: "70",
33: "75",
34: "80",
35: "85",
36: "90"
}
NoiseData = NoiseTestingImport()
TEST_ILDIPD_FeatureCON = np.empty([0,321,50,2])
TEST_ILDIPD_LabelCON = np.empty([0])
#Generate Testing Data
SpeechTestD = os.listdir('./SpeechTEST')
for FileXD in SpeechTestD:
FileX = '/SpeechTEST/' + FileXD
for SNRx in [-6,0,6]:
for Nx in range(0,len(NoiseData)):
print('Spatialising')
print('SNR: ' + str(SNRx))
print('Speech file: ' + str(FileXD))
print('Noise number: ' + str(Nx))
NoisePUT = NoiseData[Nx]
ILDIPD_Feature, ILDIPD_Label = spatialise37azimuths(FileX,SNRx,NoisePUT,Nx)
TEST_ILDIPD_FeatureCON = np.vstack([TEST_ILDIPD_FeatureCON,ILDIPD_Feature])
TEST_ILDIPD_LabelCON = np.hstack([TEST_ILDIPD_LabelCON,ILDIPD_Label.astype(int)])
######
TEST_ILDIPD_LabelCON = np.vectorize(azimuthdict.get)(TEST_ILDIPD_LabelCON)
TEST_ILDIPD_LabelCON = TEST_ILDIPD_LabelCON.astype(int)
with open('BinSL_TESTextract.npy', 'wb') as f:
np.save(f, TEST_ILDIPD_FeatureCON)
np.save(f, TEST_ILDIPD_LabelCON)
print('Genrating testing data has done!')
print('Total testing samples: ' + str(TEST_ILDIPD_FeatureCON.shape))
print('Total testing labels: ' + str(TEST_ILDIPD_LabelCON.shape))