Perceptual evaluation of order truncation in binaural Ambisonics with timbre corrections

High-order Ambisonics (HOA) encodes sound field information to varying degrees of spatial accuracy. Binaural reproduction of HOA relies on the head-related transfer function (HRTF) to avoid the need for loudspeaker arrays. There is often a mismatch of spatial resolution between the HRTF and the HOA recordings, with HRTFs having high resolutions (order 10 or above) and recordings, limited by the number of microphones, being limited to first- and second-order Ambisonics in most cases. We conducted a series of listening tests to identify the minimum truncation order for which reproduction artifacts are not noticeable. Our approach followed the MUSHRA methodology, with fifth-order recordings and HRTFs as references and null-order (no spatial information) stimuli as anchors. To focus only on the spatial richness of the stimuli after order truncation, we used timbre correction filters. Results from simple sources convolved with HOA-encoded impulse responses and HOA field recordings were presented. Both cases showed a similar trend: the presence of spatial information (first-order and higher) is perceptually meaningful; however, no statistically significant difference between second and higher orders was found. This means that proper timbre corrections can reduce the need for high-order recordings, at least when it comes to binaural reproduction.