Deuxièmes Journées Perception Sonore 2012

10 et 11 décembre 2012

CNRS Laboratoire de Mécanique et d'Acoustique

31, Chemin Joseph Aiguier

13009 Marseille


Insights from a structural theory of pitch: Existence region, sound level dependence and discriminability.


Jonathan LAUDANSKI1,2 Yi ZHENG2, Romain BRETTE2

  1. Neurelec, 2720 chemin de Saint-Bernard, Sophia-Antipolis, 06224 - Vallauris cedex,
  2. Equipe Audition, Ecole Normale Supérieure, 29 rue d'Ulm, 75230 Paris cedex 05.

Background
Pitch is a fundamental auditory percept which is perceived by many species and for humans constitutes the basis of melody production. Clearly, what distinguishes sounds with a pitch from other sounds is the temporal regularity of the signal. In other words, sounds with similar period are perceived with similar pitch. However, solely defining pitch as the perceptual correlate of periodicity raises a few discrepancies. Periodic sounds elicit pitch only within a certain domain of existence. There are small but significant level-dependencies of pitch for low frequency pure tones. Finally, although independent of the exact harmonic content (e.g. missing fundamental); the resolvability strongly determines the salience and discriminability of pitch. Here, we propose that pitch is the perceptual correlate of the similarity structure in the vibration pattern of the basilar membrane.
Methods
We use both analytical method and data analysis to study the implications behind the detection of the similarity structure. Analytical derivations are used to obtain an implicit equation linking the phase difference at two basilar membrane locations, the sound level and the estimated pitch. We analyze previously published guinea-pig auditory nerve fibers (ANF) responses to low frequency tones (Palmer and Shackelton 2009) and fit von Mises distribution to the period histogram to obtain an analytical description of the phase probability distribution.
Results
We describe the similarity structure along the basilar membrane by identifying positions whose vibrations are equal but with possible time delay. From this mathematical definition, we derive two types of structure: across channel and within channel similarity. We show that across-channel similarity is only possible for resolved harmonics while within channel similarity is possible for both resolved and unresolved harmonics. In our framework, the maximal delay allowed set the lower limit of perceived pitch and we demonstrate the implications of resolvability on this lower limit. In the case of a pure tone, the pitch shift with level is computed using phase distribution from ANF and compared to results from the psychophysics. Finally, we estimate tone discriminability as a function of F0 using vector strength.
Conclusion
Our work provides new insights into possible mechanisms used for pitch perception.