• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

"Mono correction" algorithm for BACCH

Dialectic

Major Contributor
Forum Donor
Joined
Sep 26, 2017
Messages
1,836
Likes
3,345
Location
a fortified compound
One of the (very few) quibbles about BACCH processing in the past was that it seemed to cause a subjective "thinning" of the tonality of instruments or vocals placed dead center in recordings. I particularly noticed the phenomenon on recordings of solo piano and piano concerti, but for most users, it comes up when listening to rock music, in which the lead singer and the kick drum are often located dead center. Pianos could sound treble-heavy and bass-light, and some recordings of rock music seemed to lack proper impact. On other recordings I sometimes noticed a subjective lack of control in the bass that disappeared with BACCH switched off.

I assumed that these pathologies resulted from deficiencies in the recording process, which were being ruthlessly revealed by BACCH. It turns out that that assumption was incorrect. Dead-center signals in a stereo mix can be tonally affected by crosstalk cancellation processing, even processing as sophisiticated as earlier versions of BACCH.

How did I learn this? A new update to the BACCH software is rolling out to users, and it includes a "mono correction" button with four settings: off, low, normal, and high.

After I installed the update and set mono correction to the recommended "normal" setting, the tonal thinning and other problems with center-mixed recordings were gone--just gone. Because the effect can be toggled on and off immediately, it was easy to notice the difference. Some popular music recordings (e.g., Joni Mitchell's Mingus) now have the same vocal tonality and bass drum impact that I'd hear with my Dutch & Dutch 8Cs.

The effect of the mono correction button in the software prima facie seems like a "bass boost" button, especially when listening to certain pop recordings. It becomes clear, however, that something else is going on when one listens to an exquisitely mixed stereo recording that sounds fundamentally right through prior versions of the BACCH software. (I used this.) I thought sounds located near the middle of the mix might undergo alterations in tonality with mono correction engaged (i.e., a "bass boost" for sounds that did not need it), but they did not. And somehow the bass drum hits seemed tighter and more controlled with mono correction engaged.

And crucially, for BACCH users, there seems to be no penalty in the spatial effect of BACCH from engaging the button. The same spectacular spatial retrieval of BACCH-processed music is still present with the feature engaged.

Thomas heard it a bit and seems to have liked it, too.

Another technical problem vanquished, entirely in software. My apologies for the subjective impressions above, but I don't know how to measure BACCH.
 
Thank you SoundArgument for reviewing this important new feature of BACCH4Mac. I would like to add the following explanation.

The mono correction algorithm is a result of 2 years of research aimed at solving one of the most daunting problems of crosstalk cancellation.

A mono signal in a 2-channel system is defined as one that has 100% correlation between the left and right channels. In real life it is practically impossible to produce a sound from a real acoustic source (or a speaker) with 100% correlation at the right and left ears of a listener, even if the source is equidistant from the two ears, as there will always be a finite amount of left-right de-corelation due to reflections and/or even the slightest misalignment of the head with respect to the source. However, it is very easy to produce a mono recording by having the left and right signals be the same. Any crosstalk cancellation filter, including the BACCH filter, requires a stereo signal (which by definition is not 100% L-R correlated) and therefore has a singularity when the input is mono. This singularity manifests itself as a bass rolloff that becomes more audible as the half-span between the two speakers is decreased below +/- 30 degrees. This fact can be shown mathematically. Therefore, if one plays a mono signal, or a mono-heavy signal (such as some pop music mixes where typically the lead vocalist is mixed in as a mono signal to produce a dead-center image) there would be a bass rolloff (which can be heard as a slight nasal coloration) in the sound of the center image that is audible when the BACCH filter is designed for a speaker configuration that has a speaker span (measured from the position of the listener) that is smaller than that of the regular equilateral triangle (the so-called "standard stereo triangle”). Activating the new Mono Correction algorithm completely fixes this problem. A mono signal, or the center image of a mono-heavy signal, will no longer have any coloration through a BACCH filter and will have the same tonal character as when the BACCH filter is bypassed.

For large speaker span half-angles, say 45 degrees and above, there should be no audible coloration to a mono center image through a BACCH filter and no mono correction would be needed. However, since the speakers of most stereo systems are typically configured with span half-angles smaller than that value, the new mono correction algorithm is required, and is very effective at fully restoring the tonal integrity of the center image, as you noted in your astute comments.

Buddy Gardineer
Senior Development Engineer
Theoretica Applied Physics
 
Last edited:
Back
Top Bottom