The science behind the USHER disc
Every section of the USHER disc represents the fruits of practical research, readings from the papers of other researchers and in-depth thought around the issues involved. On this page, free of the restrictions of space imposed by the CD liner notes, we give some additional insights into the rationale behind each section.
The liner notes mention the name of Alfred Tomatis. He was an audiologist who over the course of several decades developed methods of treatment, via sound, of afflictions ranging from out-of-tune singing to autism. We must stress that his methods (which are fascinating, for sure - see www.tomatis.com) are much more sophisticated than anything attempted on the USHER disc and also have different aims: we merely took some inspiration and general direction from his work.
The common thread running through Tomatis’ treatments is the presentation to the ear of sounds filtered in various ways - usually the sound of the patient’s voice. He found that this can ‘reprogramme’ the brain in quite remarkable ways. The whole business is well worth investigating, and we commend it to anyone interested in hearing and the brain.
The USHER hearing improvement tracks have a much more modest aim, and one which - although at first sight it seems far-fetched - is in fact something of an everyday occurrence. Every time you learn to recognise a new sound you have in a sense trained your hearing. But the USHER tracks go about this in a very specific way, removing the bass. Why?
All normal musical sounds consist of fundamentals and harmonics, which in terms of frequency are simple multiples of the fundamental. There may also be non-harmonic sounds, but these, like the harmonics, are concentrated in the upper frequencies. Removing the lower frequencies as we have done cuts out the fundamentals and some of the lower harmonics, but with practice the ear can quite successfully reconstruct the missing tones. When you listen to a portable radio, for instance, you are already doing this to an extent, although there the situation is generally complicated by a relatively high level of distortion.
As the ear becomes more adapted to listening to sounds with little or no low-frequency content, it is extracting more detail from the higher frequencies - it has to do this in order to maximise the ‘back-filling’ of missing frequencies. It's relatively hard work and becomes wearying after a while, like any exercise, but the effects can be surprisingly long-lasting. You'll find that as you return to versions of the recordings with more bass, up to and including the unfiltered versions, you'll still hear more detail thanks to the ear's newly-acquired sensitivity to higher frequencies. Basic measurements of hearing acuity (frequency extension and hearing thesholds) won't show up any significant change, but the ear and brain between them have become more discriminating, more efficient at extracting detail from sounds which were always well within the audible envelope but were previously under-utilised.
As a footnote to this, it's worth mentioning that most double-blind listening tests of the type carried out, for instance, to evaluate the effects of codecs such as MP3 include a significant element of ear training. The principle is broadly similar: exaggerate effects so that the ear/brain has a good chance to become accustomed to them, then reduce their size to 'normal' levels. A great number of carefully controlled academic studies have shown that listeners trained in this way are considerably more sensitive than untrained listeners.
Despite the increasing market penetration of surround sound, most people still listen to music in stereo and the vast majority of music recordings are only available in stereo. In fact stereo is capable of quite remarkable imaging in two dimensions, very nearly equal to that of full surround sound in the case of a single listener in the 'hot spot' (surround unquestionably scores when several listeners are in the room). The simple castanet tracks on the USHER disc have been included as a practical and quick way of evaluating the stereo imaging of a system. The decision to use electronically simulated stereo was taken after some experimentation with 'real' recordings, which in the end revealed as much about the inevitable deficiencies of the recording venue and equipment as about the imaging of the playback system. Using a real sound recorded with a real microphone but panned and/or delayed across the stereo image gave the best compromise between a realistic sound to which the listener can relate and accurate image placement. It guarantees, for instance, that the tonality of the sound will be the same at extreme left and extreme right, which will help track down problems with speaker tonal balance matching and room acoustics.
Note that the tonality of the sound in the centre position will not be quite the same as in the full left and full right positions. This is because high frequencies emitted by two loudspeakers are effectively decorrelated (not in a fixed phase relationship) as the speakers are several wavelengths apart and all sorts of diffraction and reflection effects will apply before the sounds arrive at the ear. At low frequencies, however, the sounds from each speaker become correlated and start to add coherently, in phase, giving a greater effective level and leading to an ultimate boost of 3dB relative to high frequencies. This is an unavoidable function of any system with more than one speaker and accounts for the quite noticeable change in tonality that many listeners have commented on in surround-sound systems between playback over all five (or more) speakers and over only the front pair in stereo.
However, there should always be symmetry between left and right, and of course the image should move smoothly across the space between the speakers. Using the USHER castanet tracks it is a fairly simple matter to adjust things like speaker spacing and toe-in, listening each time for tonal and spatial properties of the progression. Track 11 is the most useful, though tracks 13 and 15 make for a tougher test when the system is in good shape. The headphone tracks are more for interest and comparison than any such practical purpose as there's not much you can do with headphones short of upgrading to a better model!
Track 21, by the way, features time-delay increments of 5us. That's less than a sample, and provides a simple proof that digital audio systems do indeed have a time resolution that's less than the sample time (22us on CD).
It is by no means universally accepted that all hi-fi benefits from being 'burned in'. In the case of products such as cables and CD players it sounds at best questionable, and although it can be shown that many amplifiers do in fact stabilise over the first few seconds, or perhaps minutes, of operation as they reach thermal equilibrium it's still not obvious how tens or hundreds of hours of use could make a substantial difference to their performance, as many claim. In the absence of any technical evidence for this (if you have some, please let us know!) it is not possible to discount the possibility that what's really getting 'burned in' is the listener's ear, as he/she becomes accustomed to the sound of the new component and comes both to regard it as 'normal' and to extract the most information from it. Or of course there could be a combination of effects going on.
However, loudspeaker burning in is a much more prosaic matter. The suspension of the great majority of dynamic loudspeaker drivers is relatively unsophisticated and although it has been developed to the point where it works well it does not have completely stable characteristics over time. Generally, it stiffens up over the years, which makes the bass resonance of the driver rise. This has all kinds of sonic effects, including increased distortion and frequency response anomalies. USHER track 25, 'loudspeaker shakedown', simply gives the bass driver a good stretch to exercise the suspension and relieve some of its stiffness. This is completely irrelevant with full-range panel drivers such as electrostatics and may have little or no effect with some conventional (dynamic) drivers, but in general it can be surprisingly useful with many speakers. A few minutes (just put the track on repeat - we didn't want to waste space on the disc by devoting ages to something so simple) every six months is a useful general workout. Do pay attention to the preceding track for level setting as overdoing things is not a good idea. However, if you listen out for sounds of distress it's unlikely you'll come anywhere near damage levels. You can just as easily damage things by playing music too loud and we won't be held liable!
Back to the Musaeus homepage