Skip to content
A simple toolkit for speaker segmentation and identification
Find file
Latest commit 235eb8a Jun 15, 2013 @moustaki moustaki Version 0.2.1


This library provides an easy-to-use toolkit for speaker segmentation (diarization) and identification from audio.

This library is being used within the BBC R&D World Service archive prototype.

See for an example.

Speaker diarization

This library gives acccess to the algorithm developed by the LIUM for the ESTER 2 evaluation campaign and described in [Meigner2010].

It wraps a binary JAR file compiled from

Speaker identification

This library also implements an algorithm for speaker identification based on the comparison of normalised speaker models, which can be accessed through the Speaker#match method.

This algorithm builds on top of the LIUM toolkit and uses the following techniques:

  • "M-Norm" normalisation of speaker models [Ben2003]
  • The symmetric Kullback-Leibler divergence approximation described in [Do2003]
  • The detection score specified in [Ben2005]

It also includes support for speaker supervectors [Campbell2006], which can be used in combination with our ruby-lsh library for fast speaker identification.

Example use

This gem has been tested with jruby 1.7.0 onwards.

$ jruby -S gem install diarize-jruby
$ jruby -S irb
> require 'diarize'
> audio = URI('')
> audio.analyze!
> audio.segments
> audio.speakers
> audio.to_rdf
> speakers = audio.speakers
> speakers.first.gender
> speakers.first.model.mean_log_likelihood
> speakers.first.model.components.size
> audio.segments_by_speaker(speakers.first)[0].play
> audio.segments_by_speaker(speakers.first)[1].play
> ...
> speakers |= other_speakers
> Diarize::Speaker.match(speakers)

Running tests

$ jruby -S rake


[Meigner2010] S. Meignier and T. Merlin, "LIUM SpkDiarization: An Open Source Toolkit For Diarization" in Proc. CMU SPUD Workshop, March 2010, Dallas (Texas, USA)

[Ben2003] M. Ben and F. Bimbot, "D-MAP: A Distance-Normalized Map Estimation of SPeaker Models for Automatic Speaker Verification", Proceedings of ICASSP, 2003

[Do2003] M. N. Do, "Fast Approximation of Kullback-Leibler Distance for Dependence Trees and Hidden Markov Models", IEEE Signal Processing Letters, April 2003

[Ben2005] M. Ben and G. Gravier and F. Bimbot. "A model space framework for efficient speaker detection", Proceedings of INTERSPEECH, 2005

[Campbell2006] W. M. Campbell, D. E. Sturim and D. A. Reynolds, "Support vector machines using GMM supervectors for speaker verification", IEEE Signal Processing Letters, 2006, 13, 308-311

Licensing terms and authorship

See 'COPYING' and 'AUTHORS' files.

All code here, except where otherwise indicated, is licensed under the GNU Affero General Public License version 3. This license includes many restrictions. If this causes a problem, please contact us. See "AUTHORS" for contact details.

This library includes a binary JAR file from the LIUM project, which code is licensed under the GNU General Public License version 2. See for more information.

Something went wrong with that request. Please try again.