23/09/2016
HRTF: Head Related Transfer Function
This whole concept is often misunderstood as a way to EQ for stage height.
First, HRTF includes ITDs and IIDs. ITD is Interaural Time Difference. That's a fancy way of saying that a sound on one side of your body reaches one of your ears before it reaches the other one.
IID is Interaural Intensity Difference and it's a fancy way of saying that a sound on one side of your body is louder in one ear than in the other.
For really low frequencies, where the wave is MUCH longer than the distance between our ears, we don't interpret location very well. There isn't enough difference in time (phase) or the level of the wave between our left and right ears.
Above about 80Hz, level is still difficult to interpret, but there are enough degrees of phase shift for us to determine the location. That region extends up to about 800Hz, where a half a wavelenth is about the same length as the distance between our ears.
Between 800 and about 1600 Hz, we don't use phase very well and IID begins to take over. We're not good at determining location in this range--it's a transition range.
Above about 1600 Hz, ITDs are out the window and IID is how we localize sound. The sound is louder in one ear than the other.
The shape of our head and our ears are contributing factors at high frequencies because the wavelengths are too short to bend around our head and arrive at the other ear.
All of this is how we localize sound in the LATERAL plane--the means from left to right, front to back.
That's all pretty basic and it explains why delaying one speaker can help to put a phantom center image in between two speakers when we are closer to one than another. Delays are really important for midbass and midrange, but less important at high frequency-- FOR LOCALIZATION.
Having your tweeters not aligned doesn't affect localization directly, but it will affect the frequency response of the sum of the left and right speakers. Different arrival times create a comb filter--at some frequency and an odd multiples of that frequency, the sound of the two speakers are 180 degrees out of phase and that creates a null in the frequency response. At frequencies that are out of phase, we hear the speaker location rather than the sum (center).
Where understanding of HRTF breaks down is when people suggest that tuning a stereo system's frequency response according to a HRTF response measured for sound arriving from a point higher in space (overhead, for example).
The shape of our body, chin, torso, forehead and the reflections of sound that arrive from different points all aid our ability to determine the height of whatever produces sound. Sounds that originate higher than ear level are perceived with a peak in the upper midrange and a notch in the midbass. The frequency and magnitude of those peaks and dips changes based on the angle above or below eye level.
When you whistle at your dog, sometimes he c***s his head sideways and stays that way. He's listening for a second whistle so he can determine where the sound came from. This is how he uses HRTF to determine the location of the whistle.
We do the same thing. We move our heads all the time and this helps us to localize sounds to the right and left and also up and down.
When someone is mixing a movie soundtrack in multichannel and wants an airplane to sound like it flies overhead, he changes the frequency response of the plane as he pans it from the front speakers to the rear speakers. If he's recording a car moving by at eye level, he DOESN'T include that peak and notch so that the car doesn't seem to move overhead.
So...for determining the HEIGHT of a sound, HRTF works for MOVING HEADS AND MOVING SOUNDS.
How does this relate to tuning a car?
Well, I've heard people say that adding a peak and a notch or tuning for a particular vertical angle can raise the apparent height of the stage. Nope. It will appear to raise the stage for a very short time WHILE YOU'RE MAKING THE ADJUSTMENT. After the adjustment is made, your brain will incorporate the peak and the notch into the normal response and the stage will settle back down. Why? Because repeated head movements will confirm that the peak and notch are STATIC.
Our ability to detect the location from right to left for sounds in front of us is really good. From behind, it isn't so good. That's why when we suspect a sound comes from behind, we turn around. We also use our eyes. Turning around makes that work. Our ability to determine height is not as good. The difference in frequency response is greater for less acute angles. That means its easier to determine that something is overhead than it is to determine if something is a few inches higher.
