laser microphone

Reading speech off a window from outside it. Sound inside a room vibrates the glass by a few billionths of a metre; a laser or radar beam reflected off the pane carries those vibrations back, and the speech is reconstructed from them. No device is placed in the room and nothing is installed.

This is the reverse of structure-borne audio. There, a transducer drives a window so the glass speaks. Here, a beam reads a window so the glass listens. The same pane serves both, which is why it appears in accounts from both directions.

Documented real-world example: a 2026 survey in Sensors reports a millimetre-wave system recovering spoken digits at better than 93 per cent accuracy through soundproof glass at five metres, and LiDAR bounced off ordinary household objects reaching 91 per cent at 1.5 metres. Large-vocabulary speech recognition from vibration reached word error rates of 9 to 15 per cent at half a metre to 1.5 metres.

It matters for assessing a claim because it is genuinely remote — unlike every documented way of making a window produce sound, which requires physically touching it. A room can be listened to from outside without anyone entering. It cannot, on the available evidence, be spoken into the same way.

Related terms: structure-borne audio, contact microphone, piezoelectricity, image-based search

Sources: Speech recovery from vibration: laser, radar and LiDAR sensing — Sensors 26(8):2553; Protection of speech information against laser eavesdropping — Journal of Vibroengineering (2016)