[LAU] Panning based on Haas effect

Fons Adriaensen fons at linuxaudio.org
Sat Jan 19 10:48:33 UTC 2013

On Sat, Jan 19, 2013 at 12:34:57AM +0100, Jaromír Mikeš wrote:
> "R/Haas Delay" time to 0-6 ms doing nice panning C -> R
If 'panning' means 'to create an apparent source direction between
two (or more) speakers', then whatever you obtain by using such
small delays has nothing to do with the Haas effect.

Using small delays (< 0.6 ms) works well for headphones, as can
be expected - a real off-center source produces the same. It can
produce some spatial effects on a speaker system iff the speakers
are close to the listener, e.g. the typical multimedia pair sitting
besides a computer monitor. But it doesn't scale to larger setups
where most listeners can't be expected to be exactly on the center

The Haas effect is a special case of the precedence effect, which

* There is a small range of delays (from a few milliseconds up
to a few tens of ms) for which

1. the apparent source direction is that of the first sound, but
2. the delayed sound is not perceived as a separate event (echo).

The Haas effect means that this remains true even if the delayed
sound is a few dB louder that the direct one, for a somewhat smaller
range of delays.

Both cases  describe a 'snap to first' effect rather than one that
creates a virtual source between two speakers.

The precedence and Haas effects can be exploited to reinforce a sound
using speakers in a different direction without changing the apparent
direction of that sound. It can't be used for panning. 

Adding early reflections to a mix can improve the result, in some
cases dramatically. But the apparent source direction will still be
controlled by the amplitude panner used on the direct sound.



A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

More information about the Linux-audio-user mailing list