Abstract
A spectral analysis of Australian speech Is reported for 30 speakers, 15 males and 15 females. When expressed in one-third octave band levels, or critical band levels, the spectrum is essentially flat from 1 kHz to 8kHz and shows approximately 15dB greater energy at 0.5kHz and lower frequencies. Although the results of this study agree substantially with other published data, the use of the average spectrum obtained, rather than the idealised speech spectrum, would be significant for aid selection and evaluation purposes. The common practice, in hearing aid literature, of showing the speech spectrum expressed in spectrum levels, rather than one-third octave òand levels, may lead to a different, and possibly misleading, conceptualisation of the relative energy of high and low frequency components. Considering that speech spectra differ greatly from one individual to another, it would seem desirable that speech with a known, preferably average, spectrum be used in hearing aid evaluation procedures which involve setting the volume control of a hearing aid to deliver speech at a comfortable listening level.
Keywords

This publication has 12 references indexed in Scilit: