A Tutorial on Text-Independent Speaker Verification

By Frédéric Bimbot, J. Bonastre, Corinne Fredouille, Guillaume Gravier, I. Magrin-Chagnolleau, Sylvain Meignier, Téva Merlin, Javier Ortega-Garcia, D. Petrovska-Delacrétaz, Douglas A. Reynolds. 2004

954 citations recorded by the source corpus.

This paper presents an overview of a state-of-the-art text-independent speaker verification system. First, an introduction proposes a modular scheme of the training and test phases of a speaker verification system. Then, the most commonly speech parameterization used in speaker verification, namely, cepstral analysis, is detailed. Gaussian mixture modeling, which is the speaker modeling technique used in most systems, is then explained. A few speaker modeling alternatives, namely, neural networks and support vector machines, are mentioned. Normalization of scores is then explained, as this is a very important step to deal with real-world data. The evaluation of a speaker verification system is then detailed, and the detection error trade-off (DET) curve is explained. Several extensions of speaker verification are then enumerated, including speaker tracking and segmentation by speakers. Then, some applications of speaker verification are proposed, including on-site applications, remote applications, applications relative to structuring audio information, and games. Issues concerning the forensic area are then recalled, as we believe it is very important to inform people about the actual performance and limitations of speaker verification systems. This paper concludes by giving a few research trends in speaker verification for the next couple of years.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Generalized End-to-End Loss for Speaker Verification (2017, neutral): “In TD-SV, the transcript of both enrollment and verification utterances is phonetially constrained, while in TI-SV, there are no lexicon constraints on the transcript of the enrollment or verification utterances, exposing a larger variability of phonemes and utterance durations [3, 4].”
Testing for Homogeneity with Kernel Fisher Discriminant Analysis (2007, methodology): “This problem arises in many applications, ranging from computational anatomy (Grenander and Miller, 2007) to speaker segmentation (Bimbot et al., 2004).”
Kernel-Based Methods for Hypothesis Testing: A Unified View (2013, background): “Popular examples include testing for equality of signals/homogeneity, as in speaker verification [2]–[4] or change detection [5], [6].”
Privacy-Preserving Machine Learning for Speech Processing (2012, methodology): “For text-dependent tasks, the additional temporal knowledge can be integrated by using a HMM, but the use of such complex models have not been shown to provide any significant advantage over using GMMs [Bimbot et al., 2004].”
Evolving spiking neural networks for audiovisual information processing (2010, methodology): “Several ways to represent background models that can be universal or unique for each class are described and analysed in Bimbot et al. (2004).”
You Are How You Touch: User Verification on Smartphones via Tapping Behaviors (2014, methodology): “We will ask each participant to be eligible for our study if he/she plans to perform tasks on a smartphone for at least 5 minutes; (3) only raw events from smartphone sensors will be collected, and no user identities or any other user-related information will be recorded.”