On the Results of the First Mobile Biometry (MOBIO) Face and Speaker Verification Evaluation

Abstract

International audienceThis paper evaluates the performance of face and speaker verification techniques in the context of a mobile environment. The mobile environment was chosen as it provides a realistic and challenging test-bed for biometric person verification techniques to operate. For instance the audio environment is quite noisy and there is limited control over the illumination conditions and the pose of the subject for the video. To conduct this evaluation, a part of a database captured during the " Mobile Biometry " (MOBIO) European Project was used. In total there were nine participants to the evaluation who submitted a face verification system and five participants who submitted speaker verification systems. The results have shown that the best performing face and speaker verification systems obtained the same level of performance, respectively 10.9% and 10.6% of HTER

Bi-modal biometric authentication on mobile phones in challenging conditions

“…3. This challenging mobile phone database has been used to evaluate several face and speaker authentication systems [1] as well as bi-modal authentication systems [7,8]. The database provides a well defined protocol, which was initially described for the full database in [4].…”

“…An international competition was organised in 2010 [1], where researchers evaluated state-of-the-art algorithms for face and speaker authentication using Phase I of the MOBIO database [2]. In this evaluation, enrolment was exclusively performed with mobile phone data.…”

“…In particular, we examine the effect of enrolling users on high quality biometric samples acquired with a laptop computer and then authenticating them using lower quality biometric samples acquired with a mobile phone. As a significant contribution, we develop three new protocols for the MOBIO database [2] with respect to prior work [1,2,4,8] that was exclusively using mobile phone data both for enrolment and testing.…”

A Fast Parts-Based Approach to Speaker Verification Using Boosted Slice Classifiers

“…Today, speaker verification (SV) systems are gradually becoming more and more ubiquitous, finding their way into smartphones and other portable devices [1] [2] [3]. This has led to the following objectives: a) robustness against a noisy acoustic environment (additive noise) as well as channel and session variabilities, and b) computational efficiency (i.e.…”

“…Firstly, more extensive experiments were carried out to evaluate the text-independent SV performance of the system on clean speech and speech corrupted by additive noise and channel noise (using the TIMIT [16], HTIMIT [17] and noisy TIMIT databases) as well as speech data collected using mobile phones (using the MOBIO database [1] [2]…”