Spoofing and countermeasures for speaker verification

摘要: While biometric authentication has advanced significantly in recent years, evidence shows the technology can be susceptible to malicious spoofing attacks. The research community responded with dedicated countermeasures which aim detect and deflect such Even if literature that they effective, problem is far from being solved; systems remain vulnerable spoofing. Despite a growing momentum develop for automatic speaker verification, now matured sufficiently support mass deployment an array of diverse applications, greater effort will needed future ensure adequate protection against This article provides survey past work identifies priority directions future. We summarise previous studies involving impersonation, replay, speech synthesis voice conversion attacks more efforts countermeasures. should address lack standard datasets over-fitting existing specific, known