An Eyes-free User Interface Controlled by Finger Snaps

Sampo Vesa; Tapio Lokki
DAFx-2005 - Madrid
A novel way of controlling a simple user interface based on detecting and localizing finger snaps of the user is presented. The analysis method uses binaural signals recorded from the ears of the user. Transient sounds are first detected from a continuous audio stream, followed by cross-correlation based localization and simple band-energy ratio based classification. The azimuth plane around the user is divided into three sectors, each of which corresponds to one of the three “buttons” in the interface. As an example, the interface is applied for controlling the playlist of an MP3 player. The algorithm performance was evaluated using a real-world recording. While the algorithm looks promising, more research is needed before it is ready for commercial applications.