Relative importance of individual spectral features for intraconic localization

Abstract

Listen

Translate article

Most researchers agree that physical features present in the monaural spectra act as the primary set of cues for sound source localization within a cone-of-confusion. Less consensus has been reached as to whether these localization judgments are based on the presence of simple spectral features, or whether a more broad spectrum pattern matching occurs in which information across many bands are utilized. The present work first describes a head-related transfer function decomposition technique by which spectral cues are separated into components utilized for lateral localization judgments (when combined binaurally) and those used for localization judgments within a cone-of-confusion. This decomposition allows us to modify individual spectral features of an arbitrary virtual stimulus while maintaining its naturalness and perceived lateral location, criteria that were not adhered to in previous studies. Using this technique, a set of virtual localization studies was conducted in which individual spectral features were removed to examine their relative importance to intraconic localization judgments. Results indicate that while both spectral peaks and notches contribute to localization judgments, spectral information appears to be integrated across multiple frequency bands.