Review article

VOICE QUALITY AND FORENSIC SPEAKER IDENTIFICATION

Francis Nolan ; Department of Linguistics, University of Cambridge, UK

Abstract

When phonetic ians compare forensic speech samples thev o/ten remark in their reports on a "similarity of voice quality". Likewise, when eanvitnesses are asked to descrihe a voice thev have heard, thev will normally comment on the accent, if they are ah/e to, and additionally descrihe \vhat they heard as an "X voice" \vhere "X" is a temi such as "rough" or "resonant" that can he seen as an informal lahel of voice quality. In this ta/k I wi// examine these tvvo main categories of forensic speaker identification hy phonetic experts and by eanvitnesses - vvith reference to the notion of voice quality. I vvill take voice quality in the hroad sense discussed hy Laver in his T h e Phonetic D e s c r i p t i o n of V o i c e Quality (CUP, 1980), that is, as covering supralaryngeal as weU as laryngeal characteristics vvhich emerge cumulativelv from a person 's speech. In speaker comparison hy phonetic experts the emphasis in acoustic analvsis tends to he on segmenta! properties, or on pitch-related long-term features. I vvill give some examples of hovv speakers can he differentiated in this way, and touch on hovv the dynamics of formants in transitional parts of the speech signal may provide the nearest we have to a speaker's "signature". Bevond segmental analvsis, hovvever, I vvill sho\v that an analvsis using the long-term distrihutions of formant frequencies can capture information relating to Laver 's supralarvngeal voice quality categories. Given the availabilitv of Laver's comprehensive frame\vork for the impressionistic analvsis of voice quality \ve
might ask why, in the auditorv strand of their forensic analyses, phoneticians have made little use of systematic voice quality description, and / will explain why l think that is. As regards earwitness evidence / will focus on the description ofvoices by earwitnesses, and on the use of voice parades. / will ask whether an earvitness 's description of a voice might be improved if questioning of the witness vvere informed and structured by knowledge of a framevvork for voice quality description. And in creating a voice parade, / wi/l show hovv pre-tests are used to ensure that the parade is fair, including one \vhere exper i mental subjects are, in effect, asked to rate the similarity in voice quality betvveen ali pairs of samples to be used in the parade. This is to ensure that the suspect is not an outlier. Finallv I vvill previevv a project vvhich vvill investigate the effect of the telephone on such similarity judgments.

Keywords

voice quality; speaker identification; forensic phonetics

Hrčak ID:

173611

URI

https://hrcak.srce.hr/173611

Publication date:

1.9.2007.

Visits: 1.572 *