[ Home | Staff & Contacts | DIY & Tweaks | Listening tests | HiFi Playground | Music & Books ]

Memory and Judgement of Audio Quality - Part I

Criteria to Remember

[Italian version]

It is very difficult for any of us to remember sounds.
Sounds are transient phenomena.
They happen...

...then they're gone.

Human memory works by rehearsal, which is why repeated practice is needed to play music well, and repeated revision is needed to pass examinations. Our eyes are effectively an extension of our brain, unlike our ears which are separate transducers.

The repeated rehearsal is especially effective if it is conducted in different orders (of detail etc) helping to connect numerous neural pathways relating to one cluster of information. This is obviously a gross over-simplification, but you get the picture. See how visual analogy even dominates our language?

Our evolution has made visual memory dominant, indeed visual stimuli tend to be dominant in any given moment. This usually applies unless the nature of the data from each sense carries different urgency. For example, if we burn our hand on something hot we will tend not to notice the colour of the curtains at that moment. Sight often offers the opportunity for fixing its data better in our memories than any other sense because of the possibility of repeated exposure and memory rehearsal on demand.

[Ear with fuzziness]
Some ears are old
and poorly calibrated

Auditory stimuli do not offer the opportunity to replay short term impressions repeatedly to store them into our long term memory (via repetition of working memory). Music works by changing auditory stimuli over time, being even more difficult to remember than a regularly repeated short burst of sound.

Repeated playing of music through a new piece of equipment allows us to re-experience it's capabilities in various parameters of reproduction and thus learn our impressions of it. However, auditory sensation is also self adapting. Hearing adapts itself to repeated or continuous exposure of sound, hence we gradually cease to notice background hum. Thus, familiarity could reduce discernment almost as fast as it facilitates learning.

A visual analogy (our dominant sense suddenly prevailing again in this story) would be our adaptation to different colour temperatures of different light sources. Typical domestic tungsten lighting compared to workplace flourescent lighting produces a very different colour cast, but we usually fail to notice how different our clothes actually look under these different everyday lighting conditions.

Furthermore, our infant learning includes ways of filtering and prioritising stimuli. Stimuli that are less likely to be useful to us are ignored in exchange for automatic priority to that data we learn might be more useful. Room acoustics are a fine example of this in out auditory experience.
Even domestic rooms vary enormously in their acoustic properties, but we learn to discern the direct information from someone speaking because that is the data that is useful to us. We learn to ignore the myriad flutter echoes, frequency imbalances and background noise typical of most rooms. Medical conditions like brain injury and strokes indicate just how much of this automatic processing takes place, because its failure cane be very disabling & distressing.

If we walk from room to room in our own home with a friend or family member we notice the content of their conversation and the inflection in their voice before we notice the effects of the room acoustics (if we notice those at all). However, if a friend is talking to us via their telephone as they move from room to room we are much more likely to notice the changing sound of their voice.

Young ears have wider bandwidth

Seeing the nature of our surroundings generates an a priori expectation how sounds might be affected in that space. The adaptive nature of hearing and our learned response to non-auditory data combine to mitigate against the effective discernment between different pieces of audio equipment.

In order to maximise the observed difference between audio equipment under test the A-B test has become popular.

The salesman can say "listen to how X has better bass than Y" as he switches between two products. The salesman has set up an expectation of what to hear and the sudden change emphasises whatever slight difference there may be.

After a quick A-B test the reviewer may write "Y has wider and deeper soundstage than X".

Both salesman and reviewer alike are confident that the sudden change of state is really obvious. But these quick comparisons are at best superficial, and at worst misleading, as we all come to realise with experience. If it is difficult to judge, in either the short-term or the long-term, between two pieces of similar audio gear, how much more troubling is it to identify the differences and similarities between several pieces of hifi equipment, and then to remember them sufficiently to describe them accurately.

Because of this it is very difficult for us to judge between more than two different pieces of audio gear. Having spent time listening to large groups of loudspeakers (four pairs upwards) in my formative audio years I soon learned that the exercise is useless. The subjectivity of our experience and the tendency for memory to become modified by new experience renders the whole exercise invalid in any terms of qualitative data gathering.

If Not Remembered...How Reminded?

It is just possible to formulate a reasonably coherent personal narrative about the differences between two pieces of equipment and to do this with a thoroughness sufficient to render the observations as repeatable and therefore presumably reliable. Introduction of a third experience changes our recall of the earlier two experiences, they become recollected only in terms of the new experience. Some academic rigour is needed for any worthwhile presentation of subjective, therefore qualitative, data. This is a hobby, it is supposed to be fun, to divert us from the routine tasks of everyday life, so this effort seems counter-intuitive to the purpose of the exercise...that is the exercise of pleasure.

Unfortunately, to escape the uncertainties of this complexity, many audio enthusiasts still run to the seductive but illusory safety of objectivism. This objectivism is borne of scientism, the missapplication of scientific criteria to inappropriate fields or parameters. A 16oz steak is not twice as good as an 8oz steak, the weight offers us no information about how good it is at all. Furthermore both pieces of information are useless to me. I do not eat steak, and as a 21st century boy I use grammes and kilogrammes to measure mass.

Telling me that there is no weight difference (even in grammes or Newtons) between a piece of steak and a piece of unspecified other food item does not make them equally desirable or undesirable; I don't want the steak, no matter how much of it you offer and I may not want the other thing either. There are no useful measurements to communicate the subjective experience. Read the nutrition panels of a few ready-made food products and try to guess what they might taste like. Then buy some fresh ingredients instead. Slow Food is like Real Stereo. Neither can be usefully measured with simple numbers. Both should be the product of skill, craftsmanship and good ingredients. Neither should be artificially coloured nor sold through myths and magic.

Qualitative data gathering is difficult and presentation of qualitative data even more so. That is why so many audio writers begin to sound like wine writers, the task propels them into increasingly metaphor laden analogies with other (usually visual) fields. The idea of applying established qualitative data analysis like grounded theory to heuristically derived data is a non starter for ordinary audio journals and webzines, for basic reasons of time, money & expertise.

There is a very respected reviewer in the cellulose media, who frequently comes under sustained fire for attempting to allocate a cumulative points score to the subjective effectiveness of audio amplifiers. He has over 20 years experience of audio design and objective measurement, and similar time evaluating subjectively, so he should be able to manage this trick better than most writers. Amplifiers should be the most straightforward of audio components to evaluate in this way. That reviewers allocation of a numerical score should be acknowledged as his (or her) reasonable attempt to apply some kind of numeric rating to indicate any progress to the state-of-the-art offered by a landmark product. Instead its subjective basis attracts personal attacks from readers who probably couldn't understand the instruction manuals of the the conventional measuring equipment that same reviewer also uses to provide a general engineering overview of the products he tests. That cautious attempt at quantifying qualitative experience might be seen as a timid expression of the problems of expressing an emotive subjective experience in print. His attempt has attracted attacks from both sides of the audio evaluation debate.

However, at the other extreme is the alternative of opinionated tosh loaded with hyperbole 'this new mains plug transforms the humblest of systems into Krell killing wonderfulness'. This serves neither reader nor writer. It serves the writer only in their capacity to attract a brief career in the limelight, but who should soon be treated with the scepticism he (it usually is male) deserves. Indeed this serves only the advertising copyrighter who quotes it in next month's advert, or even this month's advert implying some cosy relationship not entirely in the readers interest. The cover price of most paper magazines merely covers the cost of distribution and distributors' profits. The cost of contents and the content profit contribution comes from the advertising within its pages.

So what can we do? We want to make informed decisions about our (expensive) audio purchases. The figures are generally only useful to tell us what will not work with what (you can't use a T-amp to drive those IMF TLS80s you bought on ebay). We need to establish some criteria to evaluate. We need to find some way of recording that evaluation before it passes from our brief working memory.

What do I want to hear?

Is a positive question and the easy answer is MUSIC. WE can begin to deconstruct this answer in a thoroughly post-modern narrative analysis...or we can pour a beer and put a record on. That's better.

Music is all about changes in time, so timing must be important. That is the coherence of timing between high and low notes, high and low range instruments and all the parts that make up the transient opening of a single note.

Musical time is also about rhythm. The rhythmic structure of a piece and the consistent portrayal of rhythm from one moment to the next are the fuel of music. Musical time is also about pace. Not the relentless onrush of the advancing Linnies, but the accurate pace of a good conductor or drummer.

Sometimes as a piece progresses it raises the hairs on the back of my neck. Sometimes a piece makes me tense with anticipation. Sometimes I just get lost in the enjoyment of a piece, oblivious to my surroundings. This is why I listen to music and it's called the tingle factor.

A subjective audio evaluation aide-memoire

Already there are some essential or desirable parameters we can subjectively judge in a long comparison between two pieces of apparently similar audio equipment. My personal list extends to 24 desirable audio abilities that I have identified over the years. Your list may be longer, shorter or differently prioritised. It is worth making your list because it may stop you buying initially impressive gear that doesn't really match your wants, and which therefore fails to satisfy you in the long term. This helps explain the endless treadmill of upgrades that so many of us suffer.

My list has altered only very slightly over the years, but has grown from about 8 to 24 parameters.

The 24 parameters begin with musical basics, and form 6 convenient groups:

The PRaT & emotion groupThe low-endThe midrange & vocalsThe top-endHifi attributesThe other essentials

These parameters change from item to item. What's useful to remember in a turntable comparison may be less so for a power-amplifier and vice versa. These parameters also change from music to music. The differences between two power-amplifiers will show up differently via Bartok's Six String Quartets than Captain Beefheart's Trout Mask Replica. Some samples do get chosen for their system stretching capability when whole new parameters emerge, or the balance of parameter's importance changes.

[New super ear accessory]
Dogs could be trained for
better audio evaluation

Listen to a few records or cds on your own system and make your own list. Listen to some live music, whether it's a scratch ceilidh band in a pub or a visiting symphony orchestra at a big venue, and check if the experience alters the list.

Part 2 will suggest a scheme to describe these qualities in a simple to record form using a familiar research tool adapted for the purpose.

© Copyright 2006 Mark Wheeler - www.tnt-audio.com

[ Home | Staff & Contacts | DIY & Tweaks | Listening tests | HiFi Playground | Music & Books ]