Which is the original sound?

by Helmut Weber

A vague idea of original sound

Having listened to historical recordings for more than fourty years, on Edison cylinders, gramophone records and consuming vast quantities of reissues on LPs and CDs, an idea came. The idea that it might be useful and perhaps possible to select the most originally sounding releases from the plethora of the historic output of many recording companies. Wouldn't it be fine to be able to say: this one is the best and most pleasant sounding release of a certain historic disc and its sonic impression is the same which our grandparents had of this recording? On the other hand - due to my collecting activities concerning gramophones and old record players - I knew that there might have come quite different sound to the ears of my grandparents. Anyway, it was interesting for me to make some experiments in this direction of a possible or impossible original sound, well, perhaps only a phantome.

Our test case


We have chosen a classical opera recording from the Gramophone Company's vast cataloque, one with the best Heldentenor of his time and perhaps of all times. It is Lauritz Melchior's recording of the monologue from Verdi's Otello: "Warum hast du gehäuft dieses Elend", with the New Symphony Orchestra, cond. by John Barbirolli, Electrola EJ 574, mat. CR 2510, London 1930.

Various reissues

After the end of the shellac times this classical recording has been reissued on vinyl long plays by EMI and later by other companies. Reissues today continue on CDs. We do not know, how many reissues exist. The whole spectrum is already demonstrated by those that we eventually could find in our own collection. We present here all recordings in full length together with their graphic representation by a wave picture in the audio editor. We found this representation very useful for inspecting the recordings in the large, also for endeavering some hidden trends and for an easy judgement of the noise level.

All audio samples are compressed in MP3 format with 44KHz sampling rate, 64kbps. We will also use this degree of compression for our later samples, derived from a shellac copy.

The first three reissues were made using the traditional analogue technology, but are derived of course from different copies with different qualities and use different levels or methods of restoration.

The last three reissues are digital products. The Nimbus version is unique (?) among all others because of the fact that it is recorded with a microphone from a very high quality acoustic gramophone. The History release, however, sounds quite similar to Nimbus'. If we make in-depth comparisons a suspect arises that History sells a copy of Nimbus' issue. Please compare also the two wave pictures to support this judgement.

All the others are recorded from modern turntables using magnetic or dynamic pick-ups.

Which is the best sounding and most authentic version?

Well now, which is the best reissue, or which is the most authentic one? If you have listened to our examples you know that is very difficult to answer this question. In all version a certain amount of shellac crackle is audible, at least in the piano sections. This should not be valued negatively and is certainly better than too much filtering. It is of course possible to develop a certain taste for such old electrical gramophone recordings, at least if one frequently listens to them. So someone might answer, that perhaps the Preiser release is the most pleasant one. I, for example, would say this because the Preiser LP was my first Melchior disc and I listened to Melchior's great monologue for the first time in this reissue. But this argument certainly cannot be taken too seriously. In any case from these auditions we do not know exactly which of these versions provides the most original sound.

Back to the shellac roots

So let us listen to the original shellac disc of Melchiors great recording. Well, a unique, original shellac version in most cases does not exist. In fact often there several shellac issues, from different countries or from different times. The sound qualities of such releases can differ because of different shellac qualities. In order to become more concrete we discuss here only one copy of a German shellac copy, Electrola EJ 576, which was to our disposal. It was in decent but not exceptional shape.

So again in this stadium of experiment we have a lot of choices for our listening and conversion to a digital resource. We have the choice of the (modern) turntable, pick-up, needle, pre-amplifier, equalizer, amplifier, loudspeaker and soundcard, computer system, audio-restoration software, etc. We do not speak here about more esotheric possibilities like playing 78s on an antique electric turntable or on an acoustic gramophone like the Nimbus engineers.

For the sake of concreteness I have chosen the following devices: DENON turntable, Ortophon OMB 10 cartridge, 90 micrometer needle, Onkyo receiverer, Soundblaster VVV soundcard, Adobe Audition software, Intel Pentium V, 2 GHz PC with Windows 2000 Professional.

Here are the results of our Analogue-Digital-transformation. It really sounds terribly because of much shellac crackling and many big pops. This recording needs digital restoration as it is true in nearly all similar cases. The results of subsequent restoration steps follow. Some wave pictures taken from the sceen of the sound editor optically assist the sonic impression of the samples.

As usual the greatest effect is committed by the first restoration steps, i.e. by declicking. Also reducing to true mono is a part of this task. The order in which the most important restoration steps should be applied, is well understood today. Declicking should preceed denoising and/or dehissing. But various peculiarities still are evident: whether to declick in the stereo phase or in the mono phase, how exact the equalization with respect to inverse RIAA transformation etc. should be, which denoising/dehissing algorithms should be applied, etc. So a tremendous complexity of different restoration results is possible.

Resumé

Restoration of vintage recordings is a non-trivial task which needs much experience, knowledge about performance practise and artists, and some taste. Also some engineering, physical or mathematical background is useful. The most important recipé may be the love for recordings and good connections to collectors which are able to provide clean copies. The results of audio restorations may be different, due to different technical decisions during this process and also by different initial conditions - say, quality of the copy. What we can produce by restoring old recordings begin with with the sound that modern machines can read out of the old grooves. This has to be cleaned up and annoying disturbances have to be remedied. Modern computer technology and sophisticated mathematical algorithms support this task quite well. Additionally has to be very cautious not to introduce artifacts of overrestoring.

A sort of impression of "original" sound which our ancestors might have heard in some form from a gramophone, may be generated today by playing records on such a historic device. But it is by no means optimal for today's ears and, just like restoration results, by no means unique and stays a phantome.

© H. Weber, 7 February 2012