Gyrophone: recognizing speech from gyroscope signals 论文

摘要

We show that the MEMS gyroscopes found on mod-ern smart phones are sufficiently sensitive to measure acoustic signals in the vicinity of the phone. The re-sulting signals contain only very low-frequency infor-mation (<200Hz). Nevertheless we show, using signal processing and machine learning, that this information is sufficient to identify speaker information and even parse speech. Since iOS and Android require no special per-missions to access the gyro, our results show that apps and active web content that cannot access the micro-phone can nevertheless eavesdrop on speech in the vicin-ity of the phone. 1