CTO Tomi Huttunen gave a talk at the traditional Aku-BBQ event organized by the Finnish section of AES and the Aalto Acoustics Lab. Tomi began with a short history of HRTF (head-related transfer function) simulations and how they have paved the way for today’s groundbreaking personalized HRTFs.
The first trials back in 2004 were a collaboration between the University of Kuopio and the Nokia Research Center. The original simulation tool was developed for medical ultrasonics. Though HATS (head-and-torso simulator) was a relatively crude sketch composed of simple geometric primitives, the simulation results were successfully generated over the entire audible frequency range (20 – 20000Hz).
The results showed great promise and the model was continually developed, and in 2007 HRTFs were eventually simulated for a laser-scanned model of B&K (Brüel & Kjær) HATS. The results were published in the Journal of Computational Acoustics: http://www.worldscientific.com/doi/abs/10.1142/S0218396X07003469
The original study included all of the ingredients of today’s simulation machinery, such as realistic 3-D geometries with the pinnae, or external part of our ears, of the finest detail, the full-wave solution of the physical wave equation, and the use of parallel computing to speed-up the simulations. Yet, the simulation time for a single frequency was prohibitively long.
Subsequent improvements, including faster simulation techniques, developments in cloud computing, and most important, new technologies for 3-D reality capture, were the key advancements that finally made genuinely personalized HRTF simulations possible:
And the personalization of the HRTFs make a remarkable difference! A listening test for 21 subjects shows a clear preference for personalized HRTFs over an HRTF derived from a generic dummy head. The key difference is emphasized in 3-D audio effects that have a strong directional component such as a helicopter flyover as shown below.