Assessment of changes in preference ratings of auralized engine sounds caused by changes in frequency resolution of transfer functions

Auralization facilitates aural examination of contributions from different sound sources, individually and as parts of a context. Auralizations can be created by filtering sounds of perceptually salient sources through binaural transfer functions (BTFs) from source positions to a listening position. When such auralizations are used for product sound design it is essential to know that they are of sufficient quality. A basic requirement is that preference ratings are unaffected by the quality of the auralizations. The objective of this study was to measure changes in preference ratings of auralized engine sounds caused by changes in frequency resolution of used BTFs. Auralizations of engine sounds were created by filtering source sounds through BTFs measured from source positions to a driver’s position inside a truck cabin. The BTFs were altered by lowering the frequency resolution and by smoothing in the frequency domain. Preferences for the auralizations were compared using a modified version of the MUlti Stimulus test with Hidden Reference and Anchor, MUSHRA (ITU-R BS.1534-1). Since the use of a reference is only appropriate when a reference known to be most preferred exists the reference was removed, resulting in a MUlti Stimulus Test with Hidden Anchors (MUSTHA). For assessment of the differences between the auralizations a statistical method commonly used for assessing agreement between methods of clinical measurement was adopted. The lowest frequency resolutions resulting in acceptable agreement between preference ratings of auralizations made with high frequency resolution (1 Hz) BTFs and auralizations made with simplified BTFs were 32 Hz frequency resolution or smoothing with either 1/24 octave bandwidth filters or 63 Hz absolute bandwidth filters.