UC Berkeley esearchers were able to take several 10-minute sound recordings of users typing at a keyboard, feed the audio into a computer, and use an algorithm to recover up to 96 percent of the characters entered.
The researchers used spectrum analysis, statistical learning theory, spelling and grammar checks, and learning trials to obtain recovery rates of 88 percent for words and 96 percent for characters.

The computer algorithm in the UC Berkeley study can “learn” and adapt to different typing patterns. To show this, the researchers experimented with multiple users on different keyboards, including so-called “quiet” keyboards, and found that their algorithm was successfully able to predict data. Moreover, recordings were taken in a variety of conditions, such as environments in which music was playing or cell phones were ringing in the background.

Source: UC Berkeley news release