In the preparation of historic recordings for reissue there have traditionally been two important stages of work to produce the best results:
Achieving an optimum digital transfer from the very best available source material
Restoration of the flaws in the original source's medium by means of reducing or eliminating clicks, crackles, hiss, surface noise etc. carried by that medium - tape, vinyl, shellac, acetate etc.
Until very recently the best transfers of older recordings did precisely this and no more. In fact, the better stage one went, the less one had to achieve at stage two. The philosophy of non-intervention (or at least, minimal intervention) has become quite engrained in the minds of many listeners who've previously suffered transfers which have had artificial manipulation - added reverb or echo, unrealistic stereo spread and the like - inflicted upon them.
More often than not this final treatment was done to disguise poor transfers or poor original recordings. There's little doubt that the vast majority of recordings that are classed as 'historic' have audible sonic flaws in their make-up. Many people have become used to hearing old recordings with that 'vintage' sound we associate with them and perhaps think that's how they really sounded, or should sound.
But the fact that an orchestra was recorded 70 years ago doesn't mean that the overall sound of an orchestra has really changed much in 70 years, and if you were to go back in time and listen to that orchestra in the flesh, you'd find it sounded very different to a recording of it made at the time. The fundamental relationship between root frequencies and their harmonics which gives an instrument its characteristic sound is well defined, and a 1930's clarinet should follow the same laws of physics as a 21st century clarinet in this respect.
Part 2: A New Approach - A Third Stage of Remastering
Thus we do need a third stage of remastering, and that is were XR comes in. Thus we can redefine the process as follows:
Best possible, neutral transfer
Restoration of the flaws caused by the storage medium (disc, tape etc.)
Restoration of flaws caused by the recording equipment (microphones, amplifiers, disc cutters, disc mastering etc.)
How, though, can we define what the flaws in the recording chain were, without recreating the exact recording with the same equipment? XR addresses this by working backwards from a tonal ideal, to produce a very precise equalisation curve that brings the tonal characteristics of the older recording towards that ideal. In doing this we're not adding anything to the recording that wasn't already present - instead we're compensating for the tonal defects of that recording by manipulating the recording in a precise, scientfically-deduced manner that corrects those flaws as much as is possible.
The first XR-remaster
In the development of this technique we made some important discoveries. Two or more modern recordings of the same piece of music by the same types of instruments can be harmonically analysed in such a way as to generate a sonic fingerprint which is close to being identical, regardless of the orchestra, conductor or venue - i.e. it's specific to the piece of music in question. At the same time, two recordings of different but similar works - by the same composer, same orchestra, same venue, same conductor, same recording date - will, in this method of analysis, produce quite different results - meaning the music itself holds the key.
Thus we have a means of defining the overall harmonic structure of a recording which can then be applied to another recording of the same piece. If we equalise the flawed recording to match closely the previously defined reference pattern, what we hope to achieve is stage three of the remastering process outlined above. By working backwards we've effectively deduced the tonal flaws in that recording and gone a long way towards correcting them.
This is what lies at the heart of the XR remastering system.
Putting the theory to the test
Here we take an unusual example to test the theory behind XR:
1. A mint-condition 1956 LP of Lonnie Donegan is transferred and declicked to provide our first sample. The sound is somewhat dull and constricted.
2. Next we run it through the XR process, using later recordings of the same artist to provide reference material to provide our second sample - a considerable improvement on the first, but is it accurate?
3. The third sample is a CD transfer from the master tape - sonically it's exceptionally close to the XR remaster, though lacking the extreme top and bottom end extension that XR has found.
In this case the main effect of XR has been to correct poor mastering of the original LP (I'm reliably informed that this was common at Pye at the time) and return the sound to almost perfectly match the original and well-made Joe Meek-produced masters - with a by-product of a little extra extension to the sound picture.
Conclusion: XR worked to correct sonic inaccuracies in the source recording exactly as predicted.
Part 3: Not always as straightforward as it sounds
Of course it's not always as simple as this. As soon as you start making the kind of major sonic adjustments that this technique often demands you also raise the noise floor at specific frequencies, which may then require further restoration treatment. (Sometimes it goes the other way - noise can be reduced by equalisation as well as increased, a far more effective way of dealing with it than through standard digital NR techniques).
One also finds the process trying to pull up frequencies where there's little or no signal recorded, and one has to intervene manually to eliminate this as all one hears is background noise. Therefore it's not a simple magic bullet, and XR normally creates far more work for the restoration engineer than the traditional approach, as previously hidden flaws in the medium are brought into stark focus - flaws that were not dealt with at stage two of the restoration process outline above.
At its best, however, the transformation is incredible, as an increasing number of people are discovering, both through their own listening and through often ecstatic magazine reviews.
Part 4: eXtended Range - an unexpected added bonus for many vintage recordings
And with XR we discovered there's a further and dramatic possibility. Because of the precision of the system, XR can unearth previously 'lost' harmonic frequencies, buried in the grooves of 78rpm recordings. Standard shellac grooves have yielded double their expected frequency range, from about 6kHz to over 12kHz. Furthermore, when working from original metal master parts, where the noise floor can be significantly lower, we're able to detect an even greater range of frequencies in some recordings.
This isn't always a good thing - the engineers of the day rolled off upper frequencies because their recording equipment didn't handle them too well, especially during peaks. When we bring them back up you sometimes hear why they were 'removed' in the first place - and they may need a lot of careful control and handling before they are worth hearing.
And it's not only in the electrical era that this is possible. The accepted frequency range of an acoustic 78rpm recording (pre-microphones, which started to be used in mid-1925) is roughly 250Hz-3000Hz - and yet the XR process has managed to pull out harmonics stretching right up to 9000Hz in an acoustic 1925 recording of Louis Armstrong's cornet playing; elsewhere it's made strides into the low end, rooting out near-bass notes previously lost in disc rumble.
Part 5: Further Applications of XR
The latest developments in XR revolve around its use for recordings where an identical modern equivalent neither exists nor can be recreated without great expense. Our Jazz and Blues issues suggest that it is possible to adapt XR to this end, with quite remarkable results.
Part 6: XR - a development of Pristine Audio's 'Natural Sound'
On this site you'll find some recordings classed as 'Natural Sound' and others as 'XR'. So what's the difference? Well XR grew out of 'Natural Sound' as the technique was increasingly refined to a point where it was capable of pulling out those lost upper frequencies, something which made such a difference that we felt it worth a new moniker.
XR and Natural Sound - Some Examples
Here we present a selection of recordings restored and processed with either XR or Natural Sound remastering:
Pristine Classical - DRM-free historic FLAC and MP3 downloads since 2005