2

Is there an implemented, open source, windows-compilable, speaker identification algorithm, that can be practically used for user authentication?

Bhoot
  • 2,614
  • 1
  • 19
  • 36
Maxim V. Pavlov
  • 10,303
  • 17
  • 74
  • 174
  • 2
    Any license requirements? You might also ask on http://dsp.stackexchange.com. – mtrw Dec 21 '11 at 09:24
  • @mtrw, yes, I need an open source solution. Have just added it to a question body. Thank you. – Maxim V. Pavlov Dec 21 '11 at 11:15
  • 1
    @MaximV.Pavlov: You can give bounty to the possible duplicate.You shouldn't repeat the question in the stack overflow forum. – karthik Dec 21 '11 at 12:11
  • 1
    I fixed the broken link in the duplicate. See: http://mistral.univ-avignon.fr/index_en.html – Eric Dec 21 '11 at 12:50
  • 1
    Before proceeding, I would strongly consider what your threat model is, including what attacks are feasible and how valuable the data to be protected is. Authentication using speaker identification is generally weak because a recording of the user's voice can be easily substituted for the real thing. Overcoming this requires more sophisticated techniques, or multi-mode authentication. – Chiara Coetzee Feb 05 '12 at 10:48

1 Answers1

3

There are two open source implementations for speaker identification that I know of.

  1. Modular Audio Recognition Framework
  2. ALIZE

Personally, I have worked with MARF (Java based) and it is very easy to configure and use. It works with good accuracy and comes with an implemented Speaker Identification Application which can be customized. You will get all the resources to get it working here.

Bhoot
  • 2,614
  • 1
  • 19
  • 36
  • If you have successfully implemented MARF then please help me because I am suck over it right now.It is giving me voice matched result whether I am giving correct voice sample in testing or wrong please check my question http://stackoverflow.com/questions/21037171/speaker-recognition-using-marf – Custadian Jan 10 '14 at 06:21