Search

As with any task involving statistical pattern recognition, the assessment of spoofing and anti-spoofing approaches for voice recognition calls for significantscale databases of spoofed speech signals. Depending on the application, these signals should normally reflect spoofing attacks performed prior to acquisition at the sensor or microphone. Since the collection of large quantities of any biometric data is always extremely time consuming and cost prohibitive, and consistent with some telephony applications, almost all of the existing work to assess spoofing and anti-spoofing in voice recognition is performed with databases of speech signals subjected to post-sensor spoofing attacks. While such a setup can be justified, the lack of suitable sensor-level databases is a weakness in the research field. Additionally, whereas spoofing relates to authentication, and thus predominantly text-dependent scenarios, the majority of the related work involves the use of text-independent databases better suited to surveillance applications than to authentication and spoofing.

This paper presents the first investigation of evasion and obfuscation in the context of speaker recognition surveillance and forensics. In contrast to spoofing, which aims to provoke false acceptances in authentication applications, evasion and obfuscation target detection and recognition modules in order to provoke missed detections. The paper presents our analysis of each vulnerability and the potential for countermeasures using standard NIST datasets and protocols and six different speaker recognition systems (from a standard GMM-UBM system to a state-of-the-art i-vector system). Results show that all systems are vulnerable to both evasion and obfuscation attacks and that a new generalised countermeasure shows promising detection performance. While all evasion attacks and almost all obfuscation attacks are detected in the case of this particular setup, the work nonetheless highlights the need for further research.

Progress in the development of spoofing countermeasures for automatic speaker recognition is less advanced than equivalent work related to other biometric modalities. This chapter outlines the potential for even state-of-the-art automatic speaker recognition systems to be spoofed. While the use of a multitude of different datasets, protocols and metrics complicates the meaningful comparison of different vulnerabilities, we review previous work related to impersonation, replay, speech synthesis and voice conversion spoofing attacks. The article also presents an analysis of the early work to develop spoofing countermeasures. The literature shows that there is significant potential for automatic speaker verification systems to be spoofed, that significant further work is required to develop generalised countermeasures, that there is a need for standard datasets, evaluation protocols and metrics and that greater emphasis should be placed on text-dependent scenarios.

Voice conversion is a process which converts or transforms one speaker’s voice towards that of another. The literature shows that voice conversion can be used to spoof or fool an automatic speaker verification system. State-of-the-art voice conversion algorithms can produce high-quality speech signals in real time and are capable of fooling both human listeners and automatic systems, including text-independent and text-dependent. Furthermore, since converted voice originates from a living person, traditional liveness detection countermeasures are not necessarily effective in detecting such attacks. With today’s state-of-the-art algorithms producing high-quality speech with only few indicative processing artifacts, the detection of converted voice can be especially challenging.

The vulnerability of automatic speaker verification systems to spoofing is now well accepted. While recent work has shown the potential to develop countermeasures capable of detecting spoofed speech signals, existing solutions typically function well only for specific attacks on which they are optimised. Since the exact nature of spoofing attacks can never be known in practice, there is thus a need for generalised countermeasures which can detect previously unseen spoofing attacks. This paper presents a novel countermeasure based on the analysis of speech signals using local binary patterns followed by a one-class classification approach. The new countermeasure captures differences in the spectro-temporal texture of genuine and spoofed speech, but relies only on a model of the former. We report experiments with three different approaches to spoofing and with a state-of-the-art i-vector speaker verification system which uses probabilistic linear discriminant analysis for intersession compensation. While a support vector machine classifier is tuned with examples of converted voice, it delivers reliable detection of spoofing attacks using synthesized speech and artificial signals, attacks for which it is not optimised.