-
Notifications
You must be signed in to change notification settings - Fork 32
/
conf.yaml
57 lines (39 loc) · 1.69 KB
/
conf.yaml
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
#---! THIS FILE SHOULD BE MODIFIED BASED ON YOUR CASE
#These variables should be changed
inpath: /media/anwar/D/Data/ASR/Merged_Arabic_Corpus_of_Isolated_Words/original #where input wave data exist
outpath: /media/anwar/E/Voice_Biometrics/SIDEKIT-1.3/py3env/exp/phase02 #where the program outputs exist
#training configuration
#NOTE: the summation of following sessions must be less than or equal 10
enroll_sessions: 3 #number of sessions to be included in the training set
test_sessions: 5 #number of sessions to be included in the test set
#preprocessing configurations
sampling_rate: 44100
bit_precision: 16
no_channels: 1
#features configuration (NOTE: to disable a feature, assign it to None)
features:
- vad #voice-activity-dectection
- energy #log-energy
- cep #cepstral coefficients
- fb #filter-banks
#- bnf
cepstral_coefficients: 19
filter_bank: log #filter bank can either be "log" for logarithmic and "lin" for linear
filter_bank_size: 24
lower_frequency: 300
higher_frequency: 3400
vad: snr #can be either "energy", "snr", "percentil" or "lbl".
snr_ratio: 40 #signal-to-noise ratio
window_size: 0.025 #in seconds (25 milliseconds)
window_shift: 0.010 #in seconds (10 milliseconds)
#UBM configuration
num_gaussians: 64
#i-vectors (tv) configurations (depends on UBM configurations)
batch_size: 30 #size of data batch
tv_rank: 25 # Rank of the total variability matrix
tv_iterations: 50 # number of iterations to train the variability matrix
scoring: cosine #can be either "cosine", or "mahalanobis", or "two_covariance"
#i-vectors (PLDA) configurations (depends on TV configuration)
#plotting configuration
DET_curve: rocch #you can choose either 'rocch' or 'steppy'
#other configuration