the samples of speech to be analysed were spoken by those individuals.
This unusual phonetic investigation involved the analysis of two aircraft pilots' voices following two separate incidents while flying.  In one case, the pilot was known to have become hypoxic.  In the second case, it was thought that the pilot may have become hypoxic.  My task was to analyse normal speech samples from both pilots, as well as speech samples recorded when they were suffering from, or, in the case of the second pilot, were thought to be suffering from, a lack of oxygen.
The data provided for analysis was, in one way, quite typical of most speech data provided for forensic analysis: the recordings were relatively poor in quality, with considerable extraneous noise from a variety of sources, such as the steady state noise generated by the aircraft, and the various crackles and clicks of radio transmitted signals.  Furthermore, since the data comprised recordings of transmissions of radio communication with air traffic control (ATC), a considerable amount of acoustic information was lost due to the bandwidth limitations of the radio signal.  On the other hand, the analysis differed greatly from most forensic investigations in one very important way.  Because the typical forensic case involves both known and unknown speakers, it is

(Continued on page 5)

Author: Jennifer R. Elliott, Phonetics Laboratory, School of Language Studies, The Australian National University.

Bio: Consultant and Masters Research student in forensic phonetics. B.A. in Musicology, German and Sociology. Grad. Dip. in Linguistics. (Jennifer is the recipient of the ASSTA New Researchers Award 2000.)
The term
forensic phonetics conjures up images of criminals and suspects, and the analysis of their speech, to arrive at some conclusion as to whether or not the criminal and the suspect are likely to be the same person.  A typical forensic phonetic investigation does, in fact, involve this kind of analysis.  Recently, however, I was asked to undertake a forensic phonetic investigation of speech which does not fit this model.  In this particular case, speech samples came from known speakers: there was no question as to their identity, nor was there any doubt that all