PulseAugur
LIVE 08:21:05
research · [1 source] ·
0
research

AI model recovers keystrokes with 85% accuracy using laptop microphone audio

Researchers have developed a method to recover typed text by analyzing laptop microphone audio. A convolutional neural network (CNN) was trained on log-mel spectrograms of individual keystrokes, achieving approximately 85% top-1 character accuracy on self-collected data. A PyTorch pipeline for this acoustic keystroke recovery is also available. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research highlights a potential new attack vector for sensitive data exfiltration, necessitating enhanced security measures.

RANK_REASON The cluster describes a research paper detailing a new method for acoustic keystroke recovery.

Read on Mastodon — mastodon.social →

COVERAGE [1]

  1. Mastodon — mastodon.social TIER_1 · [email protected] ·

    Acoustic keystroke recovery from laptop microphone audio. Trains a small CNN on log-mel spectrograms of individual keystrokes. ~85% top-1 character accuracy on

    Acoustic keystroke recovery from laptop microphone audio. Trains a small CNN on log-mel spectrograms of individual keystrokes. ~85% top-1 character accuracy on self-collected data, working PyTorch pipeline included. https:// pwn.guide/free/hardware/keystr oke-recovery # infosec #…