PROJECT TITLE :

Toward a Universal Synthetic Speech Spoofing Detection Using Phase Information

ABSTRACT:

In the sector of speaker verification (SV) it is nowadays possible and comparatively easy to make a artificial voice to deceive a speech driven biometric access system. This paper presents a artificial speech detector that can be connected at the front-finish or at the back-finish of a commonplace SV system, and that will shield it from spoofing attacks coming from state-of-the-art statistical Text to Speech (TTS) systems. The system described is a Gaussian Mixture Model (GMM) based binary classifier that uses natural and copy-synthesized signals obtained from the Wall Street Journal database to train the system models. Three different state-of-the-art vocoders are chosen and modeled using two sets of acoustic parameters: 1) relative section shift and 2) canonical Mel Frequency Cepstral Coefficients (MFCC) parameters, as baseline. The vocoder dependency of the system and multivocoder modeling options are completely studied. Additional section-aware vocoders are tested. Several experiments are carried out, showing that the phase-primarily based parameters perform better and are able to deal with new unknown attacks. The final evaluations, testing synthetic TTS signals obtained from the Blizzard challenge, validate our proposal.