Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Speaker verification using svm

Status
Not open for further replies.

ataufik3

Newbie level 4
Newbie level 4
Joined
Dec 25, 2012
Messages
5
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Visit site
Activity points
1,305
hi..

can anyone give suggestion for speaker verification using svm.

my target:

-programming using LabVIEW. it's posibble?
-if using svm, what are the best solution can i used?

tq
 

You should using other language, so C++,C#,...
 

In order to do a speaker verification, we need to modify the speaker data and translate it to some key data or classifiers.. these classifiers are then trained to recognise the speaker. For eg., first trim the audio sample to remove all noise utterencess and gaps between the words.. now the data we have consists only of speaker voice and no utterences and silence. Now we can normalize this signal to make it uniform for all inputs.. Now since speaker id is dependent on the frequency of the speaker's vocals, take the fft of it.. now you have two options, u can use this fft data to train the classifier or u can first pass this signal to a low pass filter to get only the speaker data as we know that human voice is band limitedd to 3.1Khz.. if u do this, u can even undersample the signal according to the filter cutoff and use that as training data. To use SVM in labview, i guess this would help
https://forums.ni.com/t5/LabVIEW/Support-Vector-Machine-Classifier-coding-in-LabVIEW/td-p/1709228

Hope this resolves your issue..
 

You need to extract formant from speech or equivalent find speaker vocal model. I think better model can be found in addition to MFCC. However you must extract features that has more discriminative value in feature space. After feature extraction learning SVM is your main problem to overcome that difficulties you must find appropriate kernel function that reduce degree of your svm function.
 
I think better model can be found in addition to MFFC. It is Speaker verification using svm.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top