Perception and production of virtual pitch by absolute-pitch musicians

• Published in: The Journal of the Acoustical Society of America Vol. 120; no. 5_Supplement; p. 3126
• Main Authors: Hsieh, I-Hui, Saberi, Kourosh
• Format: Journal Article
• Language: English
• Published: 01.11.2006
• ISSN: 0001-4966; 1520-8524
• DOI: 10.1121/1.4787676

The ability of absolute pitch (AP) and control (non-AP) musicians to identify and produce musical pitch from long-term memory was investigated in three experiments. In experiment 1, subjects identified the pitch of either a missing-fundamental complex or a highpass-filtered piano note. Unlike control subjects, AP subjects were highly accurate (\?85%) in identifying virtual pitch even when cues were restricted to the 4th and 5th harmonics. In a second experiment, subjects produced, within 3 s, the pitch of a randomly selected musical note without an external acoustic reference through either (1) adjusting a octave-range slider or (2) voicing/singing the target note. AP subjects, but not controls, were highly accurate (0.3 semitone) in the adjustment task. However, both AP and non-AP subjects were extremely accurate (i.e., SD=0.97, 0.90 semitone error, respectively) in voicing the required pitches as determined from FFTs of the recorded voiced notes. In experiment 3, subjects identified pitch of solfeggio syllables at mismatched note frequencies, created by altering the sampling rate of digitally recorded voiced labels. Identification accuracy by AP subjects declined to 65%, supporting a cognitive/linguistic interference model. Implications of these findings for models of absolute pitch encoding will be discussed. [Work supported by NSF.]