Objective Quality

"Caspar David Friedrich" in Cape Verde

The value of quality is perhaps one of the most important ones in life, and it certainly is key to people who care for music and audio. The smallest adjustment in timing can turn a nice interpretation into a breathtaking performance. Personally I have always been fascinated by how extremely small changes in an audio circuit can push the sound quality from good to addictive. 

The physical differences that cause these nuances in quality must be extremely small, but still everyone seems able to ‘feel’ it. In the seventies Robert Pirsig wrote a classic book about this matter, “Zen and the Art of Motorcycle Maintenance”. I still recommend it often to people who are struggling with judging quality and balancing the rational and romantic viewpoints on life. The core message is that you need to trust your intuition to judge quality, your consciousness should mainly be used to train your intuition and enjoy the results. 

Intuition is what your unconscious tells you. Mark that the unconscious is not born wise, it needs to be trained by experience and your rational reflection on what you experienced. Intuition can therefore be seen as your accumulated experience. The consequence is that you should only ask people for their opinion about a certain quality if you know they have experience with it. For example, do not ask people about their opinion on public transport if they always travel by car.

If you like to communicate about your quality judgement, you face a problem. First, your unconscious impression needs to be interpreted by your conscious mind to find words (and this translation is of course imperfect). Next, the other person should recognize your impression in the wording, which means you need to ‘speak the same language’. In other words: both of you should have had this experience before and linked it to the same words. This means a refined opinion can only be shared properly if both parties are experienced and have a common ‘culture’. For example, only after reading several reviews from a journalist, you get to know his writing style and experience and you can put his findings in perspective.

It gets even more difficult when you need to draw an objective verdict about a quality. In science this means that a comparison test should be done double blind and lead to statistically significant results (‘double blind’ means that no person in the room knows the right answer). In my opinion there are two important conditions to get valid results from a blind listening test. First the subject needs to clearly hear the difference between the two versions non-blind. In this phase she knows exactly what she is listening to, so her unconscious pattern recognition system is getting trained. If she does not hear a difference non-blind, it makes no sense to perform a blind test.

The second condition is special for sound: in a blind test, never play the same version twice without the subject knowing. One should never try to fool the test person. If you do, the detection threshold for sonic differences rises considerably. The reason is that a listening experience spans a certain time. By principle it is impossible to perfectly repeat that experience because environmental sounds and your mental state influence the experience and they will never be the same twice. The only correct answer when hearing the same sound again is therefore: “I hear something different”.

"Caspar David Friedrich" in Cape Verde
“Caspar David Friedrich” in Cape Verde

This has large consequences. In the traditional scientific ‘ABX test’ the assignment is to find whether A or B equals X. By principle the correct answer should always be: “In my experience A nor B is X”. If a cooperative subject does not want to break the test and tries hard to hear which stimuli are the same, her brain focusses on similarities in stead of dissimilarities. Since no two playback experiences are exactly equal, her threshold for hearing dissimilarities rises. Small differences that were clearly audible in a non-blind test therefore disappear in the ABX test.

If one likes to examine small sonic differences in a scientific way, an alternative to the ABX test is needed. We recommend a ‘trained single comparison’, or ’AAAAB’ test. In this test the subject listens to a randomly chosen stimulus A and repeats this again and again until its sonic character is stored well in her unconscious pattern recognizer. Then she switches to the alternative stimulus B and decides instantly (without switching back and forth) and intuitively whether that sounds better or worse. B should be physically different from A or the test is void. Like with ABX, the test must be repeated several times with new audio fragments. Which of the two pieces is A or B must be chosen at random otherwise the test will no longer be blind. The AAAAB test needs more trials to be of equal statistical significance as an ABX test and therefore consumes more time. But to get relevant results, it is worth the effort.

Mark that true scientific tests like this are only needed if one needs to prove a quality aspect objectively. Since making or consuming a piece of musical art is usually a largely subjective matter, it is often sufficient to trust one’s own intuition. Which is fine as long as you’re always prepared to learn and never believe you hear the objective truth. 

Eelco Grimm