I was thinking about this subject some time ago and decided to try to calculate a compensation-curve for correct phantom image projection for mono-sounds.
On http://interface.cipic.ucdavis.edu/ there is a public database of HRTF-measurements available. I first assumed a stereo-setup with an angle of 25 degrees to each speaker.
I summed the impulse-responses from the left and the right speakers (25 and -25 degrees HRTF) at the left ear (the situation is symmetric at the right ear). Then I compared this to the impulse response straight ahead (0 degree HRTF), in the frequency domain.
For the HRTF's I averaged the data for 9 of the first subjects in the database.
The resulting curve, showing the the difference in frequency response of a stereo setup compared to a pure mono source, is like this: (a compensation would be the inverse)
Much different from the compensation suggested in the thread, but it is important to take the sound from both speakers at each ear into consideration when calculating the compensation.
On http://interface.cipic.ucdavis.edu/ there is a public database of HRTF-measurements available. I first assumed a stereo-setup with an angle of 25 degrees to each speaker.
I summed the impulse-responses from the left and the right speakers (25 and -25 degrees HRTF) at the left ear (the situation is symmetric at the right ear). Then I compared this to the impulse response straight ahead (0 degree HRTF), in the frequency domain.
For the HRTF's I averaged the data for 9 of the first subjects in the database.
The resulting curve, showing the the difference in frequency response of a stereo setup compared to a pure mono source, is like this: (a compensation would be the inverse)
An externally hosted image should be here but it was not working when we last tested it.
Much different from the compensation suggested in the thread, but it is important to take the sound from both speakers at each ear into consideration when calculating the compensation.
Vg said:
Hi VG,
Thanks for the great link to the HRTF data.
I was thinking about your approach, and don't think it's quite that simple. It mixes two effects into one, but with a critical element missing.
It combines the HRTF difference due to angle, with the significant level difference due to head shadowing. However, it doesn't factor in the path delay difference in the crosstalk affect.
Given its very incomplete crosstalk cancelation, its hard to say if it makes it better or more innaccurate.
Summarizing: given left and right speakers are so far apart, given the time delay across the head, and given the head shadow affect on level, its too agressive to average -25 and +25 deg into one mono compensation.
Cheers,
Dave
DDF,
Time delay is included in the HRTF impulse response data. And through adding two impulse functions, time data is preserved when converting to the frequency domain.
I have not listened to a compensation to the curve I got or otherwise validiated it, and a "compensated" loudspeaker would surely not sound any good. I believe that compensation should be brought out in the recording process, then different compensations for different sound angles could be applied.
Time delay is included in the HRTF impulse response data. And through adding two impulse functions, time data is preserved when converting to the frequency domain.
I have not listened to a compensation to the curve I got or otherwise validiated it, and a "compensated" loudspeaker would surely not sound any good. I believe that compensation should be brought out in the recording process, then different compensations for different sound angles could be applied.
Drew Eckhardt,
That's an interesting point. The problem with correct phantom image projection should be more appearent in a damped environment with more direct-radiating speakers.
If one could do the compensation by forming the loudspeaker baffle such that the diffraction-reflections gave the correct compensation in the listening direction, the energy-curve of the loudspeaker should be unaffected.
That's an interesting point. The problem with correct phantom image projection should be more appearent in a damped environment with more direct-radiating speakers.
If one could do the compensation by forming the loudspeaker baffle such that the diffraction-reflections gave the correct compensation in the listening direction, the energy-curve of the loudspeaker should be unaffected.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.