Simulation and evaluation of a model of meter perception.
Item
-
Title
-
Simulation and evaluation of a model of meter perception.
-
Identifier
-
AAI9119658
-
identifier
-
9119658
-
Creator
-
Miller, Benjamin O.
-
Contributor
-
Adviser: Don L. Scarborough
-
Date
-
1991
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Experimental | Music | Computer Science
-
Abstract
-
A model of duration-based meter perception was developed and implemented as a computer simulation. The model, BEATS, produces a hierarchical representation of meter within the first few notes of a piece of music. BEATS was evaluated as a model of human meter perception in four experiments. The first three used the metric probe technique, in which the subject hears a very short rhythm, which is hypothesized to induce a particular meter, followed by a single probe note. The delay of the probe places it at one of several positions in the hypothesized metric hierarchy. The pattern of subjects' responses to probes in different positions served as an indication of perceived metric structure. Each experiment involved a different response: rating the probe for how well it completed the rhythm; identifying the note value corresponding to the probe's delay; and reproducing the rhythm and probe by tapping. Results from these experiments suggest that listeners do represent meter hierarchically and that, like BEATS, they can extract meter quickly and from durations alone. The fourth experiment compared BEATS analyses of longer musical rhythms with the metric structure perceived by subjects. Subjects' task was to tap the beat of the rhythm being played. By tapping with both hands subjects revealed two levels of the perceived metric hierarchy, one level to each hand. BEATS correctly predicted 40% of subjects' responses, both correct and incorrect. In addition, individual subjects tended to produce more than a single interpretation of individual stimuli, which is inconsistent with the fact that BEATS produces a single analysis of a given piece. This may be resolved by considering that (a) listeners normally extract meter not from duration alone but from several kinds of cues (including duration) at once; and that (b) when other cues are unavailable duration cues may be inadequate or ambiguous.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
