Check out this awesome yet creepy Ted talk (video embedded below or click here). Along with a bunch of other ingenious image processing technology, the video shows how researchers were able to re-create audio of a person talking by analyzing tiny vibrations in objects near the person. In one example, they shot a video of a chip bag through a window and were able to create synthetic audio of a person talking in the room via image processing alone.
This works because talking and other sounds create vibrations in the air that in turn vibrate objects in the local vicinity. These movements are extremely small and imperceivable to the naked eye, but with high definition video and the right image processing software, it’s apparently possible to detect minor changes in the video and use the data to generate synthetic audio. In many ways, this is similar to a microphone, which converts vibrations into an electrical signal, but in this case, the signal is generated visually instead of through direct connection to a vibrating substrate.
This would likely only work on certain objects or surfaces, and there would certainly be limitations based on distance of the surface from the audio source, but it’s a bit scary that it might be possible for someone to listen in on your conversations by simply recording video of objects or surfaces that are close to where you are talking.
The video below automatically starts near the end where the technology discussed above is presented, but it’s definitely worthwhile to watch the video all the way through. Enjoy.
 |
Want to know how cool technology like this is protected by patents?
|