Sound restoration: how far should it go?
SOUND RESTORATION : HOW FAR SHOULD IT GO?
The Editor invites comments from the experts in the field
Two events have had a momentous impact upon the recording scene during the past 20 years. First of all the arrival of the compact disc has resulted in what can only be described as an explosion in the availability of music at an affordable price; and secondly (and more recently) advances in digital sound technology have made it possible to improve the sound quality of pre-1950s recordings to a degree which once would have been thought impossible.
In many respects 1950 (give or take a year either way) seems to have been a watershed in sound reproduction. Until then, the 78 rpm disc had been the universal carrier of recorded music, having quickly superseded wax cylinders when sound recording became possible in the late 1800s. But two important things happened in the middle of the last century: record companies started using tape for studio recordings, and the Long Playing record arrived on the scene.
Both of these developments offered significant improvements in sound quality, both as to the quality of the actual recordings which could be faithfully captured for later replay, and in the elimination of much of the background noise which had gone hand in hand with 78s.
The familiar sound of frying eggs and bacon - the sizzle and crackle of 78s - could largely have been eliminated. In the late 1920s companies such as Columbia in Britain were issuing 78s pressed in a material which offered a much quieter surface that their contemporaries. In the last days of 78s many were pressed in vinyl (the same as LPs) and surely this could have been introduced much sooner. World War 2 was partly to blame (shellac was in short supply and recycled material wasn’t exactly pure) but also a certain amount of ‘grit’ was added at times to make 78s more durable and last longer.
Whatever the reasons (and cost must have been another consideration), the material from which most 78s were pressed was noisy, but strangely the human ear seemed able to filter it out so that listeners only heard what they wanted to hear. Also, for some perverse reason, the background noise from acoustic gramophones (with those incredibly heavy soundboxes) was less noticeable than from electric record players.
But when LPs (and to a lesser extent 45s) took over, comparisons with the older 78s highlighted the imperfections, which people regarded as no longer being acceptable.
This lengthy introduction is all leading up to the point in the early 1990s when latest developments such as the British invention CEDAR made it possible (and affordable) for record companies to start processing old recordings to extract the music, and leave the unwanted noise behind. The 50-year copyright rule in Britain also helped, because by the end of the 20th century virtually all of the 78rpm repertoire was now available to anyone who wanted to clean up and repackage vintage recordings from the past.
For the sake of accuracy, it should be mentioned that the hiss from tape recordings, and the crackles and plops from well-played LPs, can also benefit from the likes of CEDAR.
The big companies who owned the original recordings were among the first to embrace the new technology, although their enthusiasm waned as the 1990s progressed. Today it is largely the smaller independent companies who are making the running, and producing some of the best results. I suppose we should always remember that commercial companies cannot continue without profits, and their CDs of reissues must appeal to purchasers. If a touch of reverb can ‘liven up’ a dull recording and make it more appealing to record buyers, we should hardly be surprised if producers make this choice. Perhaps it is a choice between preserving a recording accurately for historical purposes, or simply making it sound enjoyable for today’s CD purchasers.
The technology is progressing at an astonishing rate. Once it could cost over £100 to process a 3-minute 78rpm disc; today anyone with a personal computer (and a little basic technical knowledge) can afford to buy the software which can process old recordings to an acceptable standard.
We can all remember the vintage LPs from the 1970s, where the method of ‘improving’ the sound quality of 78s (usually on dance band recordings) was to reduce the treble to mask the hiss and crackle. Unfortunately this removed much of the ‘bite’ in the music as well - some records sounded as though you were hearing them with a cushion stuffed inside your loudspeaker.
Today almost anything is possible ... and now we get to the real reason for this article.
The big question being asked by those with ears to appreciate the difference is: how far should today’s sound engineers go in carrying out the restoration of old records?
Should one merely try to remove some, or all of the background noise? Should a little echo be added here and there to try and ‘liven up’ dull, or dry recordings? Pre-war microphones and studios had their limitations (remember electrical recordings only arrived in the mid-1920s), and such shortcomings often become more obvious when background noise is eliminated. And on the subject of noise, should all hiss and crackle be completely removed? Sometimes this can distort the music, and result in an unreal sound from the orchestra and singers.
Through letters and comments to this and other magazines, I have become increasingly aware that record buyers are now paying far more attention to the expertise and style of the top sound engineers. Each and every one of them seem to have their strong supporters and occasionally their critics. Sometimes it can depend upon whether you normally listen on headphones or through loudspeakers. It may sound obvious, but the best sound reproduction systems can occasionally spoil listening pleasure, because any shortcomings may become more apparent.
I have therefore invited several of the top sound restorers from around the world to let me have their comments, for the benefit of readers of Journal Into Melody, and I am pleased that several have taken considerable trouble in their replies. We now print the first two responses that came in from two respected sound engineers; more will follow in March 2002.
My approach to the subject can be boiled down to a very simple rule... The goal of audio restoration should be to come as close as possible to what would have been heard if you could have stood in the studio on the day the recording was made, limitations of the recording medium itself being taken into account.
In other words, an acoustic recording will NEVER be able to sound as good as even a relatively poor electrical recording, all else being equal, simply because of the technical limitation of the two mediums. In the acoustic recording, the extended low and high frequencies of the electrical process simply do not exist on the medium, and there is no way to extract what is not there in the first place. With the limitations of the media being understood, one can then strive to reach the best that the media was capable of, consistent with my stated goal of audio restoration. The ultimate object would be to make a silk purse out of a sow's ear, but since that is not likely possible, at least, a pretty good replica of the silk purse should be attempted!
There are some modern tricks that can be employed to improve the listenability of the original material, but none of these should interfere with listening through to the original recording, to the degree that the media allows. Some of these tricks involve surgical equalization to remove resonances hums or rumble in the original. Other equalization is done to improve the sound of instruments as much as the media limitations will allow, providing the restoration engineer has a good familiarity with what live instruments should sound like.
Adding reverberation is definitely NOT permitted in the goals of audio restoration... it wasn't there in the original recording, and should not be added after the fact. Some personal advantages that I have, are a long experience in the recording field with RCA Victor, and the broadcast field with three major Canadian radio stations, both AM and FM. Additionally, doing recordings of symphony orchestras, particularly Orchestra Toronto, a 105 piece community orchestra here in Toronto. There is no substitute for being able to walk around the orchestra during rehearsals and listening to the various instruments played by excellent musicians. This gives the knowledge of the sound that you want to achieve in the restoration process.
Usually, you can focus on one instrument, a soloist perhaps, and equalize to the true sound of that instrument... if you can do that, then mostly everything else will fall in place. Granted, you may not be able to get all the overtones of the instruments, depending on the medium used to make the recording.
The last part of audio restoration is like the image in a dirty mirror... clean the mirror and the image becomes clear and sharp. Noise reduction can be thought of as cleaning the mirror, that is, removing clicks pops, crackle and hiss noises on the media that stand between you and the studio sound. The success of this part is very dependent on the recording and the media itself. The ear hears certain "cues" that tell the mind what it is listening to, and you cannot necessarily remove ALL the noise, because attempting to do so usually results in losing the natural "cues" and adding unwanted artifacts of the processing that are less desirable than the noise that was to be removed.
The experts in this field have learned this and will stop before damage is done to the music. Unfortunately, there are many so-called "producers" out there who persist in demanding that ALL the noise be removed, resulting in dull, lifeless music... and everyone has heard the sad examples of these problems... there's just no accounting for taste.
For me, the "rules of restoration" so far as (electrically recorded) 78s are concerned are quite straightforward. The aim should be to reproduce what the original recording engineer heard on the loudspeaker in the control room. This means removing the unwanted noises introduced by the recording / reproducing medium which for 78s means clicks, crackles, hiss and, in some cases, hum. Any temptations to "enhance" or "improve" the sound should be resisted although, as I will attempt to explain, it is sometimes permissible to "tweak" certain aspects of the sound which may have suffered as a result of the restoration.
Whatever certain record companies and restorers might tell you, removal of ALL of the background noise is impossible without harming the music and should never be attempted. Systems such as CEDAR used in conjunction with skilled manual editing can remove all clicks and crackles from recordings without affecting the music but, in the case of shellac 78s, usually leave behind a noise which I choose to call "shash" – a mixture of hiss and other noises caused by the shellac and whatever fillers were used in its mix. Depending on the frequencies it contains a certain amount of shash can be removed (sometimes quite a lot) using various computer processes but, as it contains a wide range of frequencies, such removal inevitably affects the music to a greater or lesser degree. This is why some highly regarded restorers do not go further than the CEDAR stage with the result that, although their work is usually more "musical" than that of others, the small amount of noise in the background means that some record company executives feel they aren’t getting value for money and insist that further "restoration" is done to remove every vestige of background noise, usually with dire results.
When even a small amount of the shash is removed, then inevitably the music suffers to some degree and it is here I feel that a judicious amount of carefully applied "tweaking" is sometimes justified. A little "presence" may be applied using parametric equalisation and, in some circumstances, a little reverberation may be added as the "ambience" of a recording is usually the first thing to suffer during restoration. However, in my opinion, this reverb must be carefully chosen to match that of the original recording and MUST be monophonic – the practice of certain record companies to swamp carefully restored recordings with stereo reverberation is, to my mind, deplorable.
There are also "restorers" who think that they can considerably improve on the original and not only add stereo reverb but carry out what I can only describe as "mutilation" of recordings by means of considerable re-equalisation which actually changes the balance between the orchestral instruments. I believe that, apart from the noise introduced by the medium, the restored sound should be as close as possible to what the original artist / producer / recording engineer intended. I consider changing it to be supreme arrogance but, as there is undoubtedly a considerable number of record buyers who like this kind of thing, this practice will undoubtedly continue and who am I to say it is wrong. However, for those of us who do not approve, I think that it should be made clear on such CDs that what they contain is not the original sound but the "what I think it should have sounded like" creation of the restoration engineer.
Although this piece is concerned primarily with the restoration of 78s, I must touch on the "restoration" of more recent mono material originally recorded on tape. These recordings rarely suffered from the limitations in the recording process which is the usual excuse offered by record companies to justify "enhancing" recordings originally made on 78s, so why some companies consider it necessary to add stereo reverberation etc. to these carefully crafted productions is beyond me. It seems impossible to justify and, in my opinion, is an insult to those who produced the original recording.
As for those who claim to be able to turn mono recordings into stereo, words fail me! The laws of physics make this impossible and anyone making such claims is not a reputable restorer but a fraud and such productions should be avoided at all costs.
At the time of writing (September) there is considerable correspondence on this topic on several e-mail groups on the Internet. One statement in particular summed it all up for me "Returning a performance to its original sound is restoration; adding "sweetening" in an effort to make it better than it is falls in the category of meddling."