Baffled Pinnae

The reason for measuring and simulating the response of baffled pinnae is mainly since the main high frequency features of HRTFs can be found in the response of these models (see Shaw, 1974).

A large baffle was constructed out of thick plywood. Its large dimensions (with a maximum width of 2.3 m, and a maximum height of 2.4 m) ensured that the measurements of the acoustical response of the pinna are not distorted due to diffraction and reflection from the boundaries of the baffle. In the centre of the baffle a circular aluminium plate was positioned, able to rotate around its centre, and a rectangular opening enables the positioning of the artificial pinnae. An additional circular plate included an opening for the holding Electret microphone, flush mounted with the baffle plane. This was used for equalisation of the transducers for each source position in space.

An external ear was moulded in the Hearing and Balance Centre of the ISVR. The moulding was created in two stages, to make sure that both the internal part of the external ear (concha, and 1 cm inside the ear canal) as well as the outside part around the pinna is captured. Two plaster models were created from the negative mould. With one was cut along the long side of the ear and included the cross section of part of the ear canal that was moulded. This cross section was required in the scanning process, to ensure all the curved parts of the pinna are captured. The frame of the plaster model was shaped and attached to an approximate rectangular frame, in a similar way to which the KEMAR pinnae are produced, so that it can be attached to the baffle in the anechoic chamber for measurements. The mesh model was decimated and mipulated to ensure accurate simulation up to 20 kHz. 

Measurements of six pinnae (four of KEMAR, one of Brüel and Kjær 4127 Head and Torso Simulator (HATS), and a replica of YK pinna) were undertaken and repeated a few times to ensure repeatability. All measurements were run automatically where a full scan of the entire sphere in a resolution of 2° took approximately one hour. A large amount of data was collected for all six pinnae, with frequency response measured at high resolution sampling of the hemisphere. 

To the mesh model of the pinna an earcanal was added. The motivation of inclusion of a simple cylindrical canal has been raised due to variations among individuals for the canal shape, size and eardrum boundary conditions. Therefore we include an average ear canal with dimensions based on the Zwislocki coupler (Zwislocki, 1970), and the boundary conditions based on average data published by Shaw (1974).