A New, and in Some Ways Preferable, A/B Comparison Technique

I have believed for decades now that the only scientifically valid technique of A/B-ing two audio components against each other was a double-blind ABX listening comparison at matched levels. Now Bill Waslo of Liberty Instruments has come up with a new methodology that has the potential of being more widely used because it is simpler, takes less time and less fussing, and is basically automated.

The essence of the process is this: Record the output of an audio component when playing any piece of music, or even a test signal. Then change something in the system—a cable, an amplifier, any other component, or apply an audiophile tweak of any kind. Next, record the output again with the same piece of music or the same test signal. Then, using Bill Waslo’s “Audio DiffMaker” software (which has not been commercially released yet but can be downloaded in a trial version), align the two recorded tracks to the same gain level and timing. Finally, subtract one from the other and listen to the difference recording. If it is basically silent, then the change has clearly done nothing. No sound in the difference recording means that the change has made no difference.  If the track is not silent, then a difference may have been made by the change, and further investigation is warranted.

A simple schematic of the process is the following, courtesy of Liberty Instruments:

Now, I am not saying (and neither does Bill Waslo) that Audio DiffMaker will convince the flat-earthers who remain unconvinced by ABX listening comparisons. There is such a thing as invincible ignorance. What I am saying is that AudioDiffmaker is scientifically sound, more convenient than ABX (automatic level matching, no endless iterations, etc.), and considerably more sensitive than the human ear. The difference recording may in some cases have a tiny signal on it below the threshold of audibility. The important thing from the audio point of view is—can you hear the difference? Listening directly to the difference, and only the difference, instead of iterated A/B comparisons is the main advantage of Audio DiffMaker. It will be interesting to see whether this new technique takes off, at least in electroacoustically sophisticated circles.

Much more detailed information is available by going to http://libinst.com/Audio%20DiffMaker.htm.