Effect of fast and slow compression on voice onset time in hearing aid processed speech

Abstract

Compression introduces distortion in hearing aid processed speech, when compared with linear amplification. Static parameters of compression namely compression ratio, compression threshold,attack time (AT), release time (RT) as well as the number of compression channels of a hearing aid may be the potential sources of compression induced distortion. The role of hearing aid compression parameters on speech perception and sound quality of hearing aids has been investigated in several studies. In some studies, it has been reported that, long ATs/ RTs leads to improvement in intelligibility of speech sounds. However, further research is required to find out the factors responsible for this. The present study therefore, investigated the influence of slow and fast compression time constants on transformation of voice onset time (VOT) in hearing aid processed speech. The objectives of the study were (1) to measure the Attack Time and Release Time of four behind the ear Digital Hearing Aids programmed in fast and slow AT/RT settings and (2) to investigate the effect of slow and fast AT/RT on VOT transformation. Six plosives in CV syllables categorized as: -unvoiced – velar /k/, dental /t/, and bilabial /p/, voiced – velar /g/, dental/d/ and bilabial /b/ produced by an adult (native Kannada) male speaker at a comfortable loudness level were recorded and used as stimulus for the study. AT and RT values of all four hearing aids in slow and fast compression setting were evaluated as per the test procedure specified by American National Standards Institute (ANSI) S3.22 standard. VOT values were measured for the input stimuli and the digitized hearing aid processed speech samples by using speech analysis software and the transformation in VOT was estimated. Results showed a reduction in VOT for voiced and unvoiced plosives at fast compression time constant setting. However, no changes in VOT was noticed at slow AT/RT setting, for unvoiced plosives, but significant difference was observed for voiced plosives.