forked from Snowdar/asv-subtools
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathivector-adapt-plda-cip-reg.py
181 lines (142 loc) · 6.31 KB
/
ivector-adapt-plda-cip-reg.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
# -*- coding:utf-8 -*-
# Copyright xmuspeech (Author: JFZhou 2020-05-31)
import numpy as np
import os
import sys
sys.path.insert(0, 'subtools/pytorch')
import libs.support.kaldi_io as kaldi_io
from plda_base import PLDA
class CORAL(object):
def __init__(self,
mean_diff_scale=1.0,
within_covar_scale=0.8,
between_covar_scale=0.8):
self.tot_weight = 0
self.mean_stats = 0
self.variance_stats = 0
self.mean_diff_scale = 1.0
self.mean_diff_scale = mean_diff_scale
self.within_covar_scale = within_covar_scale
self.between_covar_scale = between_covar_scale
def add_stats(self, weight, ivector):
ivector = np.reshape(ivector,(-1,1))
if type(self.mean_stats)==int:
self.mean_stats = np.zeros(ivector.shape)
self.variance_stats = np.zeros((ivector.shape[0],ivector.shape[0]))
self.tot_weight += weight
self.mean_stats += weight * ivector
self.variance_stats += weight * np.matmul(ivector,ivector.T)
def update_plda(self,):
dim = self.mean_stats.shape[0]
#TODO:Add assert
'''
// mean_diff of the adaptation data from the training data. We optionally add
// this to our total covariance matrix
'''
mean = (1.0 / self.tot_weight) * self.mean_stats
'''
D（x）= E[x^2]-[E(x)]^2
'''
variance = (1.0 / self.tot_weight) * self.variance_stats - np.matmul(mean,mean.T)
'''
// update the plda's mean data-member with our adaptation-data mean.
'''
mean_diff = mean - self.mean
variance += self.mean_diff_scale * np.matmul(mean_diff,mean_diff.T)
self.mean = mean
o_covariance = self.within_var + self.between_var
eigh_o, Q_o = np.linalg.eigh(o_covariance)
self.sort_svd(eigh_o, Q_o)
eigh_i, Q_i = np.linalg.eigh(variance)
self.sort_svd(eigh_i, Q_i)
EIGH_O = np.diag(eigh_o)
EIGH_I = np.diag(eigh_i)
C_o = np.matmul(np.matmul(Q_o,np.linalg.inv(np.sqrt(EIGH_O))),Q_o.T)
C_i = np.matmul(np.matmul(Q_i,np.sqrt(EIGH_I)),Q_i.T)
A = np.matmul(C_i,C_o)
S_w = np.matmul(np.matmul(A,self.within_var),A.T)
S_b = np.matmul(np.matmul(A,self.between_var),A.T)
self.between_var = S_b
self.within_var = S_w
def sort_svd(self,s, d):
for i in range(len(s)-1):
for j in range(i+1,len(s)):
if s[i] > s[j]:
s[i], s[j] = s[j], s[i]
d[i], d[j] = d[j], d[i]
def plda_read(self,plda):
with kaldi_io.open_or_fd(plda,'rb') as f:
for key,vec in kaldi_io.read_vec_flt_ark(f):
if key == 'mean':
self.mean = vec.reshape(-1,1)
self.dim = self.mean.shape[0]
elif key == 'within_var':
self.within_var = vec.reshape(self.dim, self.dim)
else:
self.between_var = vec.reshape(self.dim, self.dim)
class CIPReg(object):
"""
Reference:
Wang Q, Okabe K, Lee K A, et al. A Generalized Framework for Domain Adaptation of PLDA in Speaker Recognition[C]//ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2020: 6619-6623.
"""
def __init__(self,
interpolation_weight=0.5):
self.interpolation_weight = interpolation_weight
def interpolation(self,coral):
S_w = coral.within_var
S_b = coral.between_var
eigh_w,Q_w = np.linalg.eigh(self.within_var)
self.sort_svd(eigh_w, Q_w)
eigh_diag_w = np.linalg.inv(np.diag(np.sqrt(eigh_w)))
transform_com_w = np.matmul(eigh_diag_w,Q_w.T)
E_w,P_w = np.linalg.eigh(np.matmul(np.matmul(transform_com_w,S_w),transform_com_w.T))
B_w =np.matmul(np.matmul(Q_w,eigh_diag_w),P_w)
self.within_var = self.within_var + self.interpolation_weight* np.matmul(np.matmul(np.linalg.inv(B_w).T,np.maximum(0,np.diag(E_w)-np.eye(self.dim))),np.linalg.inv(B_w))
eigh_b,Q_b = np.linalg.eigh(self.between_var)
self.sort_svd(eigh_b, Q_b)
eigh_diag_b = np.linalg.inv(np.diag(np.sqrt(eigh_b)))
transform_com_b = np.matmul(eigh_diag_b,Q_b.T)
E_b,P_b = np.linalg.eigh(np.matmul(np.matmul(transform_com_b,S_b),transform_com_b.T))
B_b =np.matmul(np.matmul(Q_b,eigh_diag_b),P_b)
self.between_var = self.between_var + self.interpolation_weight* np.matmul(np.matmul(np.linalg.inv(B_b).T,np.maximum(0,np.diag(E_b)-np.eye(self.dim))),np.linalg.inv(B_b))
def sort_svd(self,s, d):
for i in range(len(s)-1):
for j in range(i+1,len(s)):
if s[i] > s[j]:
s[i], s[j] = s[j], s[i]
d[i], d[j] = d[j], d[i]
def plda_read(self,plda):
with kaldi_io.open_or_fd(plda,'rb') as f:
for key,vec in kaldi_io.read_vec_flt_ark(f):
if key == 'mean':
self.mean = vec.reshape(-1,1)
self.dim = self.mean.shape[0]
elif key == 'within_var':
self.within_var = vec.reshape(self.dim, self.dim)
else:
self.between_var = vec.reshape(self.dim, self.dim)
def main():
if len(sys.argv)!=5:
print('<plda-out-domain> <adapt-ivector-rspecifier> <plda-in-domain> <plda-adapt> \n',
)
sys.exit()
plda_out_domain = sys.argv[1]
train_vecs_adapt = sys.argv[2]
plda_in_domain = sys.argv[3]
plda_adapt = sys.argv[4]
coral=CORAL()
coral.plda_read(plda_out_domain)
for _,vec in kaldi_io.read_vec_flt_auto(train_vecs_adapt):
coral.add_stats(1,vec)
coral.update_plda()
cipreg=CIPReg()
cipreg.plda_read(plda_in_domain)
cipreg.interpolation(coral)
plda_new = PLDA()
plda_new.mean = cipreg.mean
plda_new.within_var = cipreg.within_var
plda_new.between_var = cipreg.between_var
plda_new.get_output()
plda_new.plda_trans_write(plda_adapt)
if __name__ == "__main__":
main()