I made a demo with some audio technique, 3D holophonic sound. I note you that this only works with headphones or at least earphones. Using the experimental data which maps the transfer functions of human auditory system, I show you real-time 3D holophonic sound effect in game is possible. Though the sound quality is quite bad since I used very rough numerical integration method, I want to emphasize that the 'feeling of existence' can be further increased not only by the visual, but also the sound.
This is the very 'prototype' and proof-of-principle demo, so the experince may not be perfect. In especially, the feeling of localized sound may be different from person to person, because the transfer functions related to the individual heads are all different. Also I only mapped the horizontal plane. This means you can only distinguish the direction of sound in the horizontal plane (Update soon..)
There are three audio samples in the demo : footstep, crunching and something shaking sound. I don't have my dev kit yet, so I need your feed back. Thanks!
It seems to work pretty well. The footsteps were the most effective for me, I think perhaps because they were the sound that was the most recognisable. The jangling thing was pretty good, but the crunching over by the female character was unnerving, I think because my brain was trying to find a possible source for the sound effect other than where my eyes were telling it it was coming from. No woman should make that noise, I think it was saying. I look forwards to more work in this direction though.
"cybereality" wrote: The stepping sound was OK, the other sounds were difficult to listen to.
Are you using a library to produce the audio or is this all custom?
sorry for the bad quality of sound. This demo was just for my curiosity and no external library was used in this demo. I just used some mathematical thing : the convolution of head-related impulse response (in time domain, experimental data) and the audio source. The convolution process involves a numerical integration, and I used very rough numerical scheme and this is the reason of the bad quality of audio. Thanks for the comment.