Fine structure analysis of speech signals. Hearing aids and perceptual training

Transition and gap detection experiments with synthetic speech stimuli in normal and pathological hearing were carried out to investigate the feasibility of some form of feature-analysing mechanism in speech perception. Identification experiments were performed before the hearing aid fitting, directly thereafter and half a year later to evaluate the important individual cues of speech perception with regard to perceptual training. Possible distortions in the temporal structure of the speech signals by the hearing aids were considered. For our experiments we generated a series of synthetic speech stimuli such as the ba-da-ga sequence, which are identical in their acoustic cues except for the transition of the second formant. The start frequency was changed in nine 200-Hz steps from 800 Hz to 2,400 Hz thus producing a ba at one end of the series and a ga at the other. The phenomenon of categorical perception could be investigated with this set of synthesized speech stimuli. In a second experiment the influence of silent intervals on the identification of plosives was analysed increasing the artificial silent interval in 10 ms steps from 0 ms to 120 ms in speech stimuli like schal thus producing stahl for the 120 ms silent interval. The results of patients with different types of hearing loss fitted with hearing aids are compared with normal hearing. Our investigations show that it will be possible to achieve an integration of fine-structure analysing mechanisms in speech perception into the management and rehabilitation of people with impaired hearing using special speech processors in modified hearing aids and developing special speech and hearing training programs.