• E.Ganesh,
• M.Prabu

Speech recognition (SR) systems like Siri or Google Now have become a popular way for humans to interact with computers, turning many systems into voice-controllable systems (VCS). However, prior research has shown that VCS can be attacked using hidden voice commands that are not understandable to humans but can still control the system. These hidden commands are audible, but we have developed a new attack called the Surfing attack, which modulates voice commands on ultrasonic frequencies, making them completely inaudible to humans. By exploiting the nonlinearity of microphone circuits, we can successfully demodulate the low-frequency audio commands, record them, and most importantly, interpret them using speech recognition systems. The Surfing attack on popular speech recognition systems was tested on Siri, Google, and Alexa. We demonstrated proof-of-concept attacks by injecting a sequence of inaudible voice commands, including activating Siri to initiate a FaceTime call on an iPhone and activating Google Now to switch the phone to airplane mode. It is also possible to detect the Surfing attack by classifying the audio using a supported vector machine (SVM) and recommend that voice-controllable systems be redesigned to be resilient to inaudible voice commands.

Surfing attack, inaudible sound, human speech, recognition.

"SURFING ATTACK DETECTION THOUGH VOICE RECOGNITION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.76-78, June-2019, Available :http://www.jetir.org/papers/JETIR1907W10.pdf

"SURFING ATTACK DETECTION THOUGH VOICE RECOGNITION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp76-78, June-2019, Available at : http://www.jetir.org/papers/JETIR1907W10.pdf