BIOACOUSTICS

Author: LOPEZ NAJERA ALBERTO.
Year: 2004.
University: SALAMANCA [www.usal.es].
Place of defense: FACULTAD DE MEDICINA.
Place of preparation: FACULTAD DE MEDICINA.

Summary: The filter nonlinear double resonance (DRNL) (Meddis et al., 2001) was designed to simulate the action of filtering that makes the basilar membrane. It has been successfully replicated the nonlinear nature of the extent of the response of the basilar membrane and the auditory nerve. There have also been reproduced data psicoacústicos related to the degree of compression applied by the human basilar membrane. However, so far it has not been proven if the filter DRNL plays other important aspects of the response of the basilar membrane such as non-linear phase or that the frequency of his impulse response is independent of sound level. Moreover, the filter DRNL is by its very structure, unable to reproduce the plateau phase and amplitude which can be observed in the experimental data for frequencies greater than the stimulation frequency characteristic of the area to register. In this paper, the filter TRNL (Triple Resonance Nonlinear) as a solution to the limitations of the filter DRNL and explores the characteristics of its response to pure tones and clicks. The structure of filter TRNL is, in essence, the filter DRNL which has been added a third branch in parallel. This is the third branch and consists of a linear filter paso-todo profit variable. The filter has a total of 15 parameters. A significant percentage of the work presented here was to achieve an optimal set of parameters to simulate the extent of the response of the basilar membrane compared with pure tones. Experimental data needed for this task were taken from the scientific literature. We considered data from seven different regions of the basilar membrane of the chinchilla with characteristic frequencies between 0.8 and 14 kHz. The input signal to the filter TRNL were previously processed through a filter that simulates the function of the middle ear. The same parameters were used to explore the response of the model compared to pure tones and face clicks. With the proper parameters, the model simulates satisfactory manner the experimental frequency response, both in amplitude and in phase, even in the region of the plateau. The response of the model in front of clicks is, however, room for improvement. While qualitatively reproduced nonlinear important aspects such as multiple lobes of the impulse response, the time it takes to achieve the maximum response, or timbreo that one sees long time. Quantitatively, the similarity between the response of the model and the pilot is poor. It is argued that the differences between the experimental and the response of the model could be because the filter used to simulate the function of the middle ear does not reflect the individual characteristics of each animal whose data were simulated. Finally, it proposes the creation of a bank of filters TRNL through regression lines that express the value of each of the filter parameters depending on the frequency characteristic.