Identification of otoacoustic emissions components by means of adaptive approximations.

Clicks and a set of tone bursts covering the same frequency band were applied as a stimuli evoking otoacoustic emissions (OAE). Recorded otoacoustic emissions were decomposed into the basic waveforms by means of high-resolution adaptive time-frequency approximation method based on the matching pursuit algorithm. The method allows for description of the signal components in terms of frequencies, time occurrences, time spans, and energy. The analysis of OAE's energy density distributions in time-frequency space revealed that click responses can be considered as linear superpositions of responses to tone bursts, The frequency-latency relationship was studied and compared with earlier works. The method made possible the exhaustive description of the resonant modes specific for given subject/ear. They were characterized not only by the close frequencies appearing for different tones, but they usually had similar latencies and time spans. Short-time and long-time resonant modes were identified. The second ones might be connected with spontaneous emissions. The method opens new perspectives in studying the fine structure of the OAE and testing of the theoretical models.