User recovery of audio operations

Abstract

Computer interfaces that support user recovery can radically alter a user's interaction style. Users can explore alternatives freely, secure in the knowledge that they can undo actions and restore previous states if necessary. A text-editor, like EMACS, where users can restore the state of an editing session to a correct previous state, is an example of such a system. Editors for textual, graphical, and many other media types commonly support user recovery. Support for and understanding of recovery in applications that use audio is not as widespread. Audio is characterized by its large volume, lack of easy indexing, and difficulty in defining inverse operations. We present a theoretical model of recovery for audio operations to help user interface designers and implementers. Our model maps an audio operation to a recovery policy and then the recovery policy to a recovery mechanism. The model uses a classification of audio operations that aids in choosing applicable recovery policies.