AWAvishai Weizman

Avishai Weizman

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Avishai Weizman has a total of 8 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Spoofing-robust speaker verification based on time-domain embedding

    by Avishai Weizman, Yehuda Ben-Shimol, Itshak Lapidot

    Spoofing-robust automatic speaker verification (SASV) technology is used to safeguard voice-based authentication systems from fraudulent attempts. Such a system should be able to verify that the speech was spoken by the target speaker, as in standard automatic speaker verification, and should also be robust against spoofing attacks. This research employs an understandable and explicable embedding derived from the probability mass function (PMF) of waveform amplitudes in the time-domain. Using the logical access (LA) database from the ASVspoof2019 challenge for evaluation, we show that the performance of the countermeasure (CM) system is enhanced when it is gender-dependent. The CM system demonstrated an equal error rate (EER) of 8.6% on the evaluation set for the male gender, and an EER of 10.1% for the female gender. In contrast, a gender-independent CM system exhibited an EER of 10.2%. As quantified by the minimum tandem detection cost function (t-DCF), the system’s performance is 0.269 for the gender-dependent CM system and 0.317 for the gender-independent system.

  • Open Access

    Recognition of Abnormal Events in Surveillance Videos using Weakly Supervised Dual-Encoder Models

    by Noam Tsfaty, Avishai Weizman, Liav Cohen, Moshe Tshuva, Yehudit Aperstein

    We address the challenge of detecting rare and diverse anomalies in surveillance videos using only video-level supervision. Our dual-backbone framework combines convolutional and transformer representations through top-k pooling, achieving 90.7% area under the curve (AUC) on the UCF-Crime dataset.

  • Open Access

    Spoofing-robust speaker verification based on time-domain embedding

    by Avishai Weizman, Yehuda Ben-Shimol, Itshak Lapidot

    Spoofing-robust automatic speaker verification technology is used to safeguard voice-based authentication systems from fraudulent attempts. Such a system should be able to verify that the speech was spoken by the target speaker, as in standard automatic speaker verification, and should also be robust against spoofing attacks. This research employs an understandable and explicable embedding derived from the probability mass function of waveform amplitudes in the time domain. Using the logical access (LA) from the ASVspoof2019 database for evaluation, we show that the performance of the countermeasure (CM) system is enhanced when it is gender dependent. The CM system demonstrated an equal error rate (EER) of 8.6% on the evaluation set for the male gender, with an EER of 10.1% for the female gender. In contrast, a gender-independent CM system exhibited an EER of 10.2%. As quantified by the detection cost function for tandem assessment (t-DCF), the system's performance is 0.269 for the gender-dependent CM system and 0.317 for the gender-independent system.


  • Open Access

    Tandem spoofing-robust automatic speaker verification based on time-domain embeddings

    by Avishai Weizman, Yehuda Ben-Shimol, Itshak Lapidot

    Spoofing-robust automatic speaker verification (SASV) systems are a crucial technology for the protection against spoofed speech. In this study, we focus on logical access attacks and introduce a novel approach to SASV tasks. A novel representation of genuine and spoofed speech is employed, based on the probability mass function (PMF) of waveform amplitudes in the time domain. This methodology generates novel time embeddings derived from the PMF of selected groups within the training set. This paper highlights the role of gender segregation and its positive impact on performance. We propose a countermeasure (CM) system that employs time-domain embeddings derived from the PMF of spoofed and genuine speech, as well as gender recognition based on male and female time-based embeddings. The method exhibits notable gender recognition capabilities, with mismatch rates of 0.94% and 1.79% for males and females, respectively. The male and female CM systems achieve an equal error rate (EER) of 8.67% and 10.12%, respectively. By integrating this approach with traditional speaker verification systems, we demonstrate improved generalization ability and tandem detection cost function evaluation using the ASVspoof2019 challenge database. Furthermore, we investigate the impact of fusing the time embedding approach with traditional CM and illustrate how this fusion enhances generalization in SASV architectures.