Localization of Sound

Sound localization is defined as the listener’s capability to recognize the origin or location of a noticed sound in distance and direction. In some cases, it’s seen as techniques in acoustical engineering to excite the placement of a hearing cue in an essential 3D space. HRTF (Head Related Transfer Function) is a reaction that characterizes how a year can receive a sound generated from a fixed point source in a given space. The head-related transfer function is normally a pair for the two years, and its primary function is to synthesize a binaural produced sound that seems to originate from a given point in a space.

HRTFs are measured in an anechoic room (a room that the walls are padded, so that very little sound enters from the outside and a very low sound reverberates). Speakers are then mounted at many locations while sounds are recorded through tiny microphones that are inserted into the auditory canals, as close to the eardrums as is possible.

The applied microphones are used to measure how much energy from diverse frequencies reaches your eardrums from different source around you. The spectrum (power and phase as a function of frequency) of the sounds source is measured and compared with the spectrum the eardrum.

Hofan, Riswick and Opsal (1998) used the same method but they had the following changes in the experiments. They inserted plastic molds into folds of adult’s pinnae to alter the elevation cue.  Four adults were then subjected to a well-fitting, custom-made shell within the concha of both years for a given period of up to six weeks. They never received any specific localization training during this period. The molds were later observed that they dramatically altered the subject’s spectral shape cues.

From the above experiments, there were many changes that occurred in the performance of the listening experience. These changes included improvements of the depicted by the systematic expansion of the responses matrices, which continued for about three weeks, after which the learning process seemed to stabilize. Hence, there was stabilization of the learning process. Next, there was the localization of accuracy steadily which improved significantly over the time for all the subjects. There was great adaptation after the twelve (PH), six (MZ), five (JO) and 29 (JR) days of continuously wearing the ear molds. Also, it resulted in reasonably, and stable, accurate localization behavior that was established in all subjects taught. Finally, it led in the control condition, immediately after the removal of the molds. All subjects were able to localized sounds with their original ears equally well as before the start of the experiment several weeks earlier.   

From the experiments, it’s clear that there is a direct relationship between pinna filtering to pinna elevation since: a broadband white noise sound will typically yield a high response for only one particular filter. Not only does this filter belong to the applied pinna set, it also corresponded to the correct target elevation. In this case, it was verified that the modified ear was able to receive specific elevation-dependent spectral features by training in a two-layer feedback neural network to map the pinna filter functions of the modified input layer onto the elevation domain (one output unit).  This is a clear illustration that there is a relationship between an individual simultaneous pinna transform functions.