Java Voice Biometric

Question

I want to develop an application based on Voice Biometric Recognition.

Specifically, I want to develop an application which will record a voice from the telephone, and identify the speaker. If the same person calls again it will recognize the voice. Like other Biometric applications do here my need is to do a voice biometric. Are there any URLs or examples which will help me. I searched but not able to find a solution.

FreeSpeech is a text-independent speaker verification system that verifies a caller's identity

I want to achieve the above one FreeSpeech Recognition in my application.

Is it possible to do the below things by using any Open Source.

1. The individual records a voice print, then
2. The system keeps track of the voice prints and can distinguish recordings from live speech

If yes, can you please provide me a URL or example which will help me.

Well, I got the light from This Url to achieve the above task but not able to get the expected out put.

Solution 1

After wasting 20 to 25 Hrs, Finally I got the solution by using MARF Framework. I got the sample app from the http://sourceforge.net/projects/marf/files/Applications/%5Bf%5D%20SpeakerIdentApp/0.3.0-devel-20060226/ And for now, it's working fine for me. This links is very useful for me to make the sample app executable. http://marf.sourceforge.net/

Solution 2

You can take a look at this previous SO post in which various Java Speech Recognition Engines are described such as Sphinx.

Solution 3

I am not an expert on this domain so please take my answer as is , it's not an authorative one... I think you have different ways to achieve your goals : - finding a Java library is one , the most natural one - recording the voice in Java then applying one of the several algorithms available for such job , you may find many research papers dealing with that subject - depending from the architecture choices, you may find different libraries implemented in C dealing with voice signal, using JNI or JNA is one way to deal with C/C++ libraries, Web Services or CORBA are other ways to achieve this....

HTH Jerome

Comments

[subodh]

Thanks @npinti, do you have any experience on this?.

[npinti]

@subodh No, I have never used any speech engine. You should be able to find something online though.

[npinti]

I do not know if it is possible or not, like I have said, I have never used such a framework. I think that the second point might be hard to achieve. Maybe you can take a look at some signal processing material.

[Piotr Gwiazda]

I've one used SciLab behind Java application to do cepstral analysis of signal to check some biometrics from human voice.

[amaurycrickx]

For the record, here is another Java based speaker recognition framework named Recognito which can be integrated in a matter of minutes instead of hours : github.com/amaurycrickx/recognito It didn't exist at the time this question was raised but it might be useful to people checking today...