Subject: Re: : Sound Source Segregation and head motion From: Peter Lennox <P.Lennox(at)DERBY.AC.UK> Date: Mon, 19 Dec 2005 14:07:03 +0000On this question of head movements. several questions /observations: 1) Why conceptually separate "head movements" and whole-body movements? 2) Is it necessary to assume that rotation is the key ingredient in head movement? 3) Since binaural hearing is always found in ambulant perceivers and never in static organisms (e.g plants) might we suppose that the mobile component is as valuable for auditory spatial perception as are the binaural difference components? 4) In any event, taking an ecological argument that 'sources' rarely appear in anechoic circumstances, and so spatial hearing has evolved to usually deal with perceptual tasks in echoic environments, are head-movements about cognitively mapping 'what-and-where'. - That is, 'where-features' (in a sound field) such as occlusions, reflective bodies, semi-enclosed space resonances, etc., will never physically behave as point sources and so are easily disambiguated from actual sources during successive movements. Thus, movement could be part of a primitive sorting mechanism to distinguish 'objects' (:sources) and 'features' and 'overall place' prior to mapping..? 5) In this view, rotating sources is equivalent to head rotation ONLY in the special laboratory circumstances where there is no background 'ambience labelling' (I've previously toyed with the idea that this ambience labelling is in some ways equivalent to the 'ground' in discussions of figure-ground separation in visual thinking). Therefore, one would, in order to investigate whether they were equivalent in real environemntal terms, have to rotate an entire environment. Physically, this would mean modelling the impulse response of an entire place in which sources find themselves, and rotate that. This could be done, and my (uninformed) guess is that, if it were, the effect would be disorientating and would lead to interesting adaptation effects. 6) Doppler effects are less to do with rotation than 'passing'; simply rotating a source around the listener shouldn't produce any, though there would be doppler-affected reflections in an echoic environment. regards ppl Dr. Peter Lennox S.P.A.R.G. Signal Processing Applications Research Group University of Derby http://sparg.derby.ac.uk Int. tel: 1775 >>> Pierre Divenyi <pdivenyi(at)EBIRE.ORG> 17/12/2005 16:35 >>> Head motion temporarily changes a stationary source into a moving object. In the early 90's, Chandler and Grantham showed that the detectable angle of a moving object is inversely proportional to its bandwidth. In a theoretical paper** we showed that this trade-off is a property of physics -- in other words, motion changes spectral resolution and vice-versa. Consequently, by moving his/her head the listener can introduce subtle spectral cues that could be potentially useful for disambiguating spectral grouping of one or more simultaneously present sources. Pierre Divenyi **Divenyi, P. L. & Zakarauskas, P. (1992). "The effect of bandwidth on auditory localization: An estimation theory model", Auditory Physiology and Perception, edited by Cazals, Y., Demany, L. & Horner, K., (Pergamon Press, London) pp. 563-570. ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________