| Autor: |
Evsyukov, Michael, Putyato, Michael, Makaryan, Alexander |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2700 Issue 1, p1-6, 6p |
| Abstrakt: |
The article reviews the existing approaches used in speaker verification systems and explains why rapid development of voice authentication is taking place at the moment. The peculiarities of voice authentication systems in comparison with other biometric methods are described. The general scheme of the voice authentication system is provided. Metrics for assessing the effectiveness of voice authentication are presented. The characteristics of the most commonly used discriminative voice features are described. The classification of the most popular algorithms for automatic speaker verification is presented. The most popular spoofing methods and their essential features are characterized. The varieties of antispoofing countermeasures are described and their efficiency at protecting from different kinds of spoofing is specified. A modified classification of antispoofing countermeasures is proposed. The research indicates upcoming study directions in the speaker authentication sphere. Modern speaker verification methods are widely used and reputed. One of their major advantages is that they can be adapted not only for one-time, but also for continuous authentication. In spite of that, voice authentication systems are not completely reliable. Low price and high quality audio processing tools make them vulnerable to spoofing. Therefore, new measures aimed at preventing, detecting and protecting from spoofing should be invented to assure information security. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
