Speaker-specific information in voice quality parameters

Abstract

Seven instances of the vowel /a/, embedded in a carrier sentence with a variety of prosodic and segmental conditions and each produced by twenty male German speakers were measured acoustically for a variety of supralaryngeal, laryngeal, and sublaryngeal types of voice quality, as well as for fundamental frequency. Measurements included parameters that reflect the variation of the voice source in the dimension between modal and breathy phonation, as well as the influence of subglottal pressure on the voice source. The voice source parameters were corrected for supralaryngeal influence according to a procedure proposed by Kenneth N. Stevens and students, arriving at parameters such as H1 * - H2*. Post hoc comparisons between speakers revealed that the parameters that carry most speaker-specific information are F3, F0, and H1 * - H2 •. The value and limits of the results for the forensic task of speaker identification are discussed.