The instrumented audio-source object allows an application to record the audio passing through to the recognizer, or replace the audio going to the recognizer with audio from a wave file. The object can be used for speech recognition.