APerceptualModelofPulseSalienceandMetricalAccentin.pptVIP

A Perceptual Model of Pulse Salience and Metrical Accent in Musical Rhythms Richard Parncutt Presented by Zhenyao Mo Overview Learning patterns (meters) Locating accents (downbeat) Two experiments A model Perceptual Quantitative Computational Experiment-1 : Pulse Salience Six Patterns Pulse (1) Waltz (2:1) March (2:1:1) Swing (3:2:1) Skip (3:1:2) Cross (2:1:1:2) Tempi 6 tempi Total : 6 * 6 = 36 trial 22 Listeners To tap along with the underlying beat Data Pattern Tempi Period (time between taps) Phase (start of the rhythm) Experiment-1 : Conclusion Pulse sensation is quite ambiguous Ambiguity increases as the pattern becomes more complex Ambiguity increases as note rate increases Small increase inconsistent Experiment-2 : Metrical Accent Six Patterns Pulse (1) Waltz (2:1) March (2:1:1) Swing (3:2:1) Skip (3:1:2) Cross (2:1:1:2) Tempi Only 1 Along with basic patterns, target notes were played in another timbre Total : 26 trial 22 Listeners To judge whether the target was Off the beats (sure) Off the beats (not sure) On the beats (not sure) On the beats (sure) Conclusion: ambiguous Model : overview Model (continue) Model (continue) Model (continue) Model (continue) Model (continue) Model (continue) Model (continue) Application of the Model Read the paper … * * Duration in ms Saturation function Phenomenal accent Pattern matching Pulse Salience Tempo Dependence Max Tactus Superposition Metrical Accent L.H.’s Model : begin at the start of a rhythm, then test hypothetical interpretations At each new event Present Model : look at a whole sequence at once Duration in ms Saturation function Phenomenal accent Pattern matching Pulse Salience Tempo Dependence Max Tactus Superposition Metrical Accent Either absolute time Or relative to a basic time Say, b = ? note beat Cycle: one duration of basic time Duration in ms Saturation function Phenomenal accent Pattern matching Pulse Salience Tempo Dependence Max Tactus Superposition Metrical Accent Each cycle is assign a