METHOD FOR DETECTING AN AUDIO ADVERSARIAL ATTACK WITH RESPECT TO A VOICE COMMAND PROCESSED BY AN AUTOMATIC SPEECH RECOGNITION SYSTEM, CORRESPONDING DEVICE, COMPUTER PROGRAM PRODUCT AND COMPUTER-READABLE CARRIER MEDIUM

摘要

The disclosure relates to a method and device for detecting an audio adversarial attack with respect to a voice command (VC) processed by an automatic speech recognition system (ASR). The method is implemented by a detection device connected to the automatic speech recognition system. The method includes: obtaining (11) an audio signal associated with the voice command; performing (12) a phonetic transcription of the audio signal, according to a phonetic transcription scheme, delivering a first character string (CS1); obtaining (13) a transcript resulting from the processing, by the automatic speech recognition system, of the audio signal; performing (14) a phonetic transcription of the transcript, according to the phonetic transcription scheme, delivering a second character string (CS2); computing (15) a similarity score (SS) between the first character string (CS1) and the second character string (CS2); and delivering (16) a piece of data representative of a detection of an audio adversarial attack, as a function of a result of a comparison between the similarity score (SS) and a predetermined threshold.