Improving speaker verification performance under spoofing attacks by fusion of different operational modes

Saeid Safavi, Hock Gan, Iosif Mporas

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    6 Citations (Scopus)

    Abstract

    In this paper, we propose a methodology for the fusion of different modes of speaker verification (SV) operation (fixed-passphrase, text-dependent and text-independent mode), using regression fusion models. The experimental results with and without spoofing attack conditions and using different single mode speaker verification engines, GMM-UBM, HMM-UBM and i-vector, indicated improvement in all the experiments. The 6.75 % in terms of EER is achieved as the best speaker verification performance, when using fusion of scores from three modes of operation of HMM-UBM based speaker verification systems. Relative improvement of 22.32 % achieved compare to the best performing single mode engine.

    Original languageEnglish
    Title of host publicationProceedings - 2017 IEEE 13th International Colloquium on Signal Processing and its Applications, CSPA 2017
    PublisherInstitute of Electrical and Electronics Engineers (IEEE)
    Pages219-223
    Number of pages5
    ISBN (Electronic)9781509011841
    DOIs
    Publication statusPublished - 10 Oct 2017
    Event13th IEEE International Colloquium on Signal Processing and its Applications, CSPA 2017 - Penang, Malaysia
    Duration: 10 Mar 201712 Mar 2017

    Publication series

    NameProceedings - 2017 IEEE 13th International Colloquium on Signal Processing and its Applications, CSPA 2017

    Conference

    Conference13th IEEE International Colloquium on Signal Processing and its Applications, CSPA 2017
    Country/TerritoryMalaysia
    CityPenang
    Period10/03/1712/03/17

    Keywords

    • anti-spoofing
    • Automatic speaker verification
    • GMM-UBM
    • HMM-UBM
    • i-vector
    • MLP
    • regression fusion
    • spoofing attack
    • SVM

    Fingerprint

    Dive into the research topics of 'Improving speaker verification performance under spoofing attacks by fusion of different operational modes'. Together they form a unique fingerprint.

    Cite this