Abstract
Investigations into frequency analysis in the ear have been made from various points of view, and reports by many researchers show that the ear exhibits different characteristics according to the conditions of stimulus presentation because of the neural complexity. Almost all of the sounds which we experience in everyday life suffer changes in contents, for instances, durations of vowels and consonants are usually less than 100 msec, and also in musical sounds a steady state does not even continue for a second. To know the auditory response to such a sound, therefore, it is necessary to study the temporal process of frequency analysis in the ear. In this article we discuss by means of temporal masking how the auditory excitation pattern along the frequency axis is developed when a tone burst or a short tone containing two frequency components is presented, and further consider the time required for frequency analysis in the ear on the basis of detection threhelds of frequency transition of sounds with the initial steady segment and the final one. When a signal tone (with duration of 30 msec) is given close on the heels of a masker (tone burst of a frequency of 1, 000 Hz), the difference between the threshold shift at 1, 000 Hz and ones at other frequencies (1050 and 1100 Hz) is nearly zero for masker's duration of 10 msec, and increases as duration of the masker is lengthened to attain a stationary value at about 100 msec (Fig. 2). Investigating into the wider frequency range of signal sound, the masking curves (masking audiograms) are obtained as shown in Fig. 3 (a) and (b). No peak aspears on the masking curve for masker's duration of 10 msec. This is considered to show the imperfection of lateral inhibition in the neural network. It must be noted that the threshold shift at 1, 000 Hz for the masker's intensity of 95 dB SPL (Fig. 3-a) is a little different from that for the intensity of 80 dB SPL (Fig. 2), when the duration of masker is 10 msec. For masker's duration of 25 msec, a peak is clearly observed at 1, 000 Hz. When a complex tone with two components (with frequencies of 800 and 1, 000 Hz) is given as a masker, a duration of more than 25 msec is required for the appearance of the peaks at frequencies of the components (Fig. 4). Frequency width (-3dB band width) of the aural response to a single frequency tone (with a frequency of 1, 000 Hz) calculated from the above results decreases along with an increase of duration of the tone, and finally attains the constant value of about 15 Hz at the duration of 100〜300 msec. This result means that the band width of the auditory filter changes as a function of time, and it is translated into a change of its sharpness in Fig. 13. When the frequency of a tone burst is linearly changed from the initial frequency of 1, 000 Hz during a portion of the burst, detection thresholds of rapid frequency transition expressed in the frequency difference between its initial frequency and the final one depend on the duration of a steady part before and after the transition, and attain a constant value for the steady duration of 100〜300 msec (Fig. 9 and Fig. 11). For the shorter transition duration (5 msec), the relation between the duration of a steady part before the transition and the detection threshold corresponds approximately to the relation between duration of tone burst and the one-fifth frequency-bandwidth of aural response (Fig. 12). When the duration of a steady part is short, the detection threshold of frequency transition is affected by the transition duration a lot and the longer the transition duration, the easier it is to detect the transition. This range may require consideration from other angles such as the behavior of neurons sensible to frequency modulated tone.
