facial features. The offender may have worn a mask, blindfolded the witness, committed the offending
exclusively over the telephone, or the witness may have been sight-impaired. In these situations,
witnesses might still be able to use their other senses to identify the offender.
This article will discuss
witnesses that rely on their aural sense to identify an offender, particularly using voice identification.
The witness in this instance is not an eyewitness, but is instead an earwitness (Yarmey, 2012).
Judicial treatment of earwitness testimony over the last century has been disparate; some courts have
given it more credence than it deserves,
others have unfairly dismissed it.
Nevertheless, courts in
England, the United States, Canada and Australia have now generally accepted that a defendant can be
identified solely by their voice (Ormerod, 2001). Indeed, in most Australian jurisdictions, voice iden-
tification of a defendant is specifically included in legislation as a permissible form of evidence to prove
that the defendant is the person who committed the crime in question.
There are five types of voice identification evidence that may be adduced in criminal proceedings:
1. recordedsamples of the offender’svoice for comparisonwith the voice of the defendantby the jury;
2. evidence of a non-expert that a voice heard on a particular occasion was that of the offender
(either recognising the voice during the offending or identifying it subsequent to the offending);
3. evidence by ad hoc experts on the similarity between two voices. Their ad hoc expertise comes
not from any scientific training but from repeated listening to a particular offender’s voice and
then comparing it to the voice of the defendant (such as can occur when police officers or
translators listen to many hours of listening device recordings of a particular person);
1. For example, witnesses could identify an offender by reference to the senses of touch (e.g. through a deformity or texture of the
skin), smell or taste (e.g. the smell of cannabis on the offender’s breath); AK vWestern Australia  HCA 8 at .
2. Mack vState 44 So 706 (Fla. 1907) (‘The manner, time and place of his assault upon [the victim] threw her instantly into the
highest state of terror and alarm, when all of her senses and faculties were at the extreme of alert receptiveness, when there was
nothing within her reach by which to identify her assailant but his voice. Who can deny that under these circumstances that
voice so indelibly and vividly photographed itself upon the sensitive plate of her memory, as they she could forever afterwards
promptly and unerringly recognize it on hearings its tones again’).
3. RvPaxton  1 VR 178 (‘if visual identification is to be scrutinized carefully, identification by voice only must be
scrutinized even more carefully. There are, of course, cases where the conviction has rested on voice identification, but in every
case identification of that nature must be regarded in my opinion as weak evidence’); O’Donnell vRowe and Smith 
NTMC 42 (a magistrate finding a voice identification unreliable because the magistrate did not personally find the respondent’s
voice distinctive, because in the magistrate’s experience voices over the phone are different to voices in person, because the
magistrate did not believe the applicant had sufficient occasion to learn the respondent’s voice, and because the magistrate
considered the applicant’s refusal to concede possible error to make her identification less reliable.)
4. Evidence (National Uniform Legislation) Act (NT) Dictionary; Evidence Act 2011 (ACT) Dictionary; Evidence Act 2008 (Vic)
Sch 2; Evidence Act 2001 (Tas) s. 3; Evidence Act 1995 (NSW) Dictionary; Evidence Act 1995 (Cth) Dictionary. It is also at
least arguable that South Australia permits the admissibility of voice identification evidence with the double-negative language
in section 34AB of the Evidence Act 1929 (SA) (‘In a criminal trial, evidence of the identity of a person alleged to have
committed an offence is not inadmissible...merely because it was obtained other than by means of an identity parade’)
5. Hodgson (2007: 6). The list provided in the present article differs from Justice Hodgson’s in two ways; first, he includes a
category of experts testifying about how reliable evidence from the other categories might be (which we do not); second, he
does not include voice-matching technology evidence in his list.
6. Jury voice comparisons have been deemed an acceptable form of voice identification evidence; Bulejcik vThe Queen, (1996)
185 CLR 375
. See also RvSolomon  SASC 265 at  (‘I proceed on the basis that Bulejcik is authority for the
proposition that the jury can compare recordings of voices to decide for themselves whether or not the voice on one recording is
the same as the voice on another recording’); Korgbara vThe Queen  NSWCCA 84.
7. Or even asserting that a voice was from someone of a particular gender; see, e.g., RvRatten  VicSC 264 (a telephone
operator’s testimony that a call was made by a woman requesting emergency assistance from police was admissible as proof
against the defendant’s assertion that he was the one calling for emergency assistance from an ambulance).
voice identifications in Australia, see Edmond and San Roque (2009: 21–27); Edmond et al. (2011a).
McGorrery and McMahon 263