Time-frequency methods for studying non-stationary auditory responses

The study of temporal processing of acoustic sig-nals by the auditory system becomes increasingly important for the development of new strategies for diagnosis and treat-ment of people who are deaf or hearing impaired. Objectives: To evaluate three time-frequency methods (Short Time Fouri-er Transform, Morlet Wavelet Transform, and a newly pro-posed Chirp Analyzer) for the reliable estimation of non-stationary auditory electrophysiological responses, which could be used in the study/diagnosis of hearing problems in clinical practice. Methods: Using simulated and real data, we compare the robustness and reliability of the three methods for different levels of noise and response forms, as well as with different physiological response delay. Results: In general, the three methods provide a fairly reliable estimate of the physio-logical response when there are low levels of noise and re-sponse latency is small. The Chirp Analyzer is faster and more robust to noise, while continuous Morlet Wavelet Transform is very sensitive to noise but more reliable when the responses appears with a considerable delay. Conclusions: Results sug-gest that the Chirp Analyzer is a promising tool for estimating non-stationary auditory electrophysiological responses, alt-hough future validation is needed.