Non-sibilant fricatives in English: spectral information above 10 kHz

The purpose of this study is to explore the contribution of higher frequencies to fricative consonant classification in Australian English. The spectral range above 10 kHz is analysed in an attempt to provide better classifications of the dental and the labiodental fricatives. A comparison is made between real-word data and data from a controlled-word context. Although results for the controlled-word context suggest that spectral information above 10 kHz may be of use in classifying [f] correctly, these results are not repeated for the real-word data. Moreover, classification of [f] and [q] using only the information above 10 kHz yields quite poor results. It is also shown that the results for the voiced dental and labiodental fricatives are not as promising as those for their voiceless counterparts.