Header image  
by Ralph Glasgal
line decor
Home Tutorials Tech
Kudos and
Demos Bio Free Ambio
Glossary The Home
Concert Hall
Rec Engineers
FAQ/Forum Links Contact us
line decor


One Key to Ambiophonic crosstalk cancellation
ITDspkr inter-aural time difference from a positioned loudspeaker

by Robert E. (Robin) Miller ©2008

Crosstalk cancellation (XTC), by which speakers can act like headphones to reproduce life-like binaural audio, is used in Ambiophonics. The difference between Ambiophonics and most past methods is that, rather than angling speakers 60° as with stereo, the speakers are together in front, angled 10~20°. For important central sounds, this arrangement obviates stereo’s problems of comb filtering and pinna angle confusion while it alters localization. It eliminates stereo’s shifting of images toward the speakers that create the hole-in-the-middle for offside or too-close listeners.. Correcting stereo replay with Ambiophonics works because sound from the speakers comes from where the central images are.

Crosstalk is inevitable using speakers. Delay artifacts
(when sounds travel farther to the opposite ear) reduce
clarity and distort center voice timbre with comb-filtering.

In any XTC method, the signal of one channel is delayed, attenuated, inverted, and mixed with the other speaker’s signal so that it arrives at its ear in time to cancel the crosstalk from the first speaker. For one or two listeners on the median line bisecting the speakers, this XTC signal cancels the crosstalk acoustically, leaving each ear to hear only the speaker on the same side, i.e. with no crosstalk. (It is the equivalent of acoustically isolating each speaker to its ear, such as with a barrier.)

Various methods have been used over the years to generate XTC signals, ranging from building into each speaker a second driver recessed and fed the opposite channel, to DSP that convolves the inverse of crosstalk delay, attenuation, inversion, and mixing recursively within PC audio host, in VST plug-ins, or in firmware within audio hardware, such as first manufactured by TacT.


Key to XTC is the delay that occurs when sound from speaker L rounds the head to ear R, and vv. This inter-aural transit time difference varies with the angle between speakers and the listeners. The four variables are: speed of sound (c, varies with temperature & humidity), ear separation (inter-aural distance IAD, varies by individual), speaker distance, and speaker separation. The last two form an equilateral triangle for conventional 60° stereo, but the angle is smaller for PC or TV speakers, or an “Ambiodipole” speaker pair.

As an example, for 68°F room temperature and 50% relative humidity, c=344.6 m/s. The equivalent average spacing between a listener’s ears (inter-aural distance IAD) was established by Blauert as 0.175m, the diameter D of a head-equivalent sphere. Under these conditions, conventional stereo crosstalk for speakers angled 60° (e.g. separated 1.80m, positioned 1.80m from the listening position) results in ITDspkr of 260µs – the delay inherent in all crosstalk signals. For a center phantom (including important soloists or dialogue), this time lag produces nearly identical but 260µs-delayed signals that mix acoustically at each ear, causing comb filtering (raspy dips in frequency response beginning about 2kHz and up) and smearing transient details (reduces clarity). The brain interprets these duplicates as extra early reflections that come from the speakers, not anything recorded. Sounds that are just off-center are “relocated” toward the speakers, creating the hole-in-the-middle. Prior XTC methods such as Lexicon and Carver retain this 60° geometry, along with similar distortions for important central voices.

Ambiophonics on the other hand locates speakers close together, for example separated 0.5m at the same 1.80m distance from the listener (a resulting angle of 16.0°) with an ITDspkr of 71µs. Setting this delay using RACE XTC in DSP, TacT, or a VST plug-in, the listener will experience with most stereo recordings no center image problems, a very wide 120° to 180° stage – perceived outside the speakers – and linear imaging side-to-side. (Compare this with stereo’s limitation to the 60° width between speakers). The downside of Ambiophonics is that with an imprecise layout, or listening off-axis, RACE XTC signals will not cancel at the ears, and a mono signal will be heard. Contrast this with offside stereo or 5.1 where only one channel is clearly audible.

Ambiophonics positions two speakers closely in front
and uses crosstalk-cancellation DSP. PanAmbio adds a
second pair in back for 360° surround (5.1-compatible).


For placement geometries differing from the above illustration, the following formula may be pasted in a spreadsheet with XTC’s four values in cells as below. Calculated ITDspkr is: =1000000*H3/2*(ASIN((F7/2)/D7)+((F7/2)/D7))/D3 where: D7 is distance from listener to speakers, in m; F7 is the separation between speakers, in m; D3 is the speed of sound, in m/s (typ. 344.6); H3 is the ear spacing, in m (avg. 0.175). Calculated ITDspkr for typical media room layouts will range between 40~179µs (185µs for typical PC/gaming speakers, 114µs a typical laptop). Furthermore, the angle between speakers measured from the center of the listener’s head will be =2*57.3*ASIN((F7/2)/D7), ranging 9~41° (42° for PC speakers, 26° for a laptop).