Control of voice fundamental frequency in speaking versus singing

Ulrich Natke, Thomas M. Donath, and Karl Th. Kalveram

The Journal of the Acoustical Society of America -- March 2003 -- Volume 113, Issue 3, pp. 1587-1593

Institute of Experimental Psychology, Heinrich-Heine-University, Düsseldorf, Germany

Abstract:

In order to investigate control of voice fundamental frequency (F0) in speaking and singing, 24 adults had to utter the nonsense word ['ta:tatas] repeatedly, while in selected trials their auditory feedback was frequency-shifted by 100 cents downwards. In the speaking condition the target speech rate and prosodic pattern were indicated by a rhythmic sequence made of white noise. In the singing condition the sequence consisted of piano notes, and subjects were instructed to match the pitch of the notes. In both conditions a response in voice F0 begins with a latency of about 150 ms. As predicted, response magnitude is greater in the singing condition (66 cents) than in the speaking condition (47 cents). Furthermore the singing condition seems to prolong the after-effect which is a continuation of the response in trials after the frequency shift. In the singing condition, response magnitude and the ability to match the target F0 correlate significantly. Results support the view that in speaking voice F0 is monitored mainly supra-segmentally and controlled less tightly than in singing.
(Bold text emphasis by Martin Braun)

Comment:

The results of this study show for the first time that the precision of automatic pitch control in speech is related to a basic musical ability. In recent years automatic intonation control in the speaking voice had repeatedly been measured by the technique of pitch-shifted auditory feedback. If speakers hear their own voice only via headphones and if it is then pitch-shifted by electronic manipulation, a reflex-like and subconscious compensation sets in. An unnoticed down-shift in the feedback leads to an equally unnoticed compensatory up-shift in the voice, and vice versa. Thus it was demonstrated that speakers have internal representations of pitch targets.
The results of Natke et al. now permitted a comparison of pitch-shift compensation in speech with the musical ability of pitch imitation in singing. Subjects were asked to sing a presented piano tone that had been selected from the speakers normal pitch range for speech. From those subjects for which a compensatory pitch-shift response in speech could be registered, 16 sang the piano tone with a mean deviation of < 100 Cent (one semitone). Nine of these 16 even sang with a mean deviation of < 30 Cent, thus showing the basic musical ability to imitate a tone within safe limits of its pitch-class (chroma). The other seven subjects from the group of 16 sang with mean deviations of 35 to 84 Cent.
Then, in the test of pitch-shift compensation in speech, the more musical group of 9 subjects showed clearly larger pitch-shift compensations (mean 63 %, range 42-117 %) than the less musical group of 7 subjects (mean 44 %, range 21-80 %). Thus the more musical group had a more precise automatic control for expressing internally represented pitch targets in speech. This new finding once again indicates that the neural mechanisms for speech and music may be more closely related than is usually assumed.
(Comment Martin Braun)

NOM Home