Item

Abstract

Morphing gradually from one stimulus to another is useful for studying whether subjects perceive stimulus transitions gradually or categorically. Signal processing algorithms are now available for morphing human speech sounds, which had been more difficult to achieve than for visual objects. These techniques would be useful to implement for preparing ethologically relevant stimuli for behavioral and neurophysiological study with animal models, like the macaque (Macaca) species of monkeys. However, morphing complex sounds requires developing methods for selecting landmarks to guide the morphing process, which is difficult for classes of vocalizations that differ on many acoustical parameters. We describe a procedure for using the freely available STRAIGHT signal processing package to morph between: 1) harmonic lsquo;coolsquo; vocalizations from two macaques, 2) different types of vocalizations from the same individual, a lsquo;cool
squo; and a //!-
-MFG_und--//lsquo;gruntlsquo;, and 3) monkey and human vocalizations. We evaluated the quality of the morphs and obtained classification curves from human listeners who seemed to categorize the monkey vocalizations like those produced by humans. The procedure prepares macaque vocalizations for neuroethological study and the methods should also be useful for successful morphing between different classes of animal and human vocalizations.