Logo des Repositoriums

Auflistung nach Autor:in "Kasper, Klaus"

1 - 3 von 3
  • Konferenzbeitrag
    Continuous speaker verification in realtime
    (BIOSIG 2011 – Proceedings of the Biometrics Special Interest Group, 2011) Kunz, Max; Kasper, Klaus; Reininger, Herbert; Möbius, Manuel; Ohms, Jonathan
    Biometric speaker verification deals with the recognition of voice and speech features to reliably identify a user and to offer him a comfortable alternative to knowledge-based authentication methods like passwords. As more and more personal data is saved on smartphones and other mobile devices, their security is in the focus of recent applications. Continuous Speaker Verification during smartphone phone calls offers a convenient way to improve the protection of these sensitive data. This paper describes an approach to realize a system for continuous speaker verification during an ongoing phone call. The aim of this research was to investigate the feasibility of such a system by creating a prototype. This prototype shows how it is possible to use existing technologies for speaker verification and speech recognition to compute segments of a continuous audio signal in real-time. In line with experiments, a simulation study was made in which 14 subjects first trained the system with a freely spoken text and then verified themselves afterwards. Ad- ditional intruder tests against all other profiles where also simulated.
  • Konferenzbeitrag
    Efficient two-stage speaker identification based on universal background models
    (BIOSIG 2014, 2014) Billeb, Stefan; Rathgeb, Christian; Buschbeck, Michael; Reininger, Herbert; Kasper, Klaus
    Conventional speaker identification systems are already field-proven with respect to recognition accuracy. Since any biometric identification requires exhaustive 1 : N comparisons for identifying a biometric probe, comparison time frequently dominates the overall computational workload, preventing the system from being executed in real-time. In this paper we propose a computational efficient two-stage speaker identification system based on Gaussian Mixture Model and Universal Background Model. Binarized voice biometric templates are utilized to pre-screen a large database and thereby reduce the required amount of full comparisons to a fraction of the total. Experimental evaluations demonstrate that the proposed system is capable of significantly accelerating the response-time of the system and, at the same time, identification performance is maintained, confirming the soundness of the scheme.
  • Konferenzbeitrag
    Improving channel robustness in text-independent speaker verification using adaptive virtual cohort models
    (BIOSIG 2012, 2012) Nautsch, Andreas; Schönwandt, Anne; Kasper, Klaus; Reininger, Herbert; Wagner, Martin
    In speaker verification, score normalization methods are a common practice to gain better performance and robustness. One kind of score normalization is cohort normalization, which uses information about the score behaviour of known impostors. During enrolment, impostor verifications are simulated to get a speaker-specific set of the most competitive impostors (the cohort). In the present paper, one virtual cohort speaker is synthesized using the most competitive impostor's Hidden Markov Models (HMMs). These impostors are also users of the system and therefore their models have channel-specific information contrary to the universal background model, which provides channeland speaker-independent models. On verification, cohort scores are obtained by an additional verification of the virtual cohort speaker. The cohort scores evaluate the candidate as an impostor. A cohort normalized score promises greater robustness. This paper will study the effect of the introduced cohort normalization technique on the speaker verification system atip VoxGuard, which is based on mel-frequency cepstral coefficients and HMMs. VoxGuard can be used as either a text-dependent or a text-independent verification system. In this paper, emphasis is placed on textindependent speaker verification. Experiments using the atip speech corpus and the SieTill speech corpus showed improvements measured by the equal error rate on performance and robustness.