Who hasn’t dreamt of hearing Bach play his works at the organ, or improvise on the fortepiano, as he did for the King of Prussia? Who wouldn’t have liked to attend a Chopin piano recital (Frederic, not Henri)? The quest for the preservation of traces of the past goes way back: the exceptionally vivid Fayum mummy portraits, dating from the 1st century AD, are one of its many manifestations.
But music works, and music performance even more so, are a different proposition. Music notation probably started over 4000 years ago on cuneiform tablets (which were also the medium on which library catalogues first appeared).
Mechanical music production is much older than the 18th century automata: about the year 265 BC, an engineer called Ctesebius invented or improved the water organ (hydraulis). Later developments allowed the organ to be played automatically: a cylindrical drum, revolving by means of a water wheel, had projecting pieces fixed to its curved parts, almost like those of more modern carillon players. When each projection comes into contact with a balanced lever, a hole on a horizontal pipe is uncovered and the note sounded by this pipe is heard1: music automata existed already 2000 years ago.
But music reproduction had to wait until the invention of Charles Cros in 1877 – one year before Edison – of a “procedure for the recording and the reproduction of audible phenomena”. We all know how it has affected – and continues to do so – the reception of music by the public.
In 1904, Edwin Welte invents the Mignon, designed to record not the sound produced by the piano, but the action of the pianist – and thus to allow for a much better reproduction of the original performance: this is why we can listen to a very clean rendition of Beethoven’s Ecossaise in E flat as played in 1905 by Carl Reinecke and recorded in 2006 on a Steinway Welte. This was a momentous landmark in the pursuit of perfect preservation.
But as this device worked only for the piano (and necessitated to own the instrument to listen to the recording), most of the performances of the past – including those for piano only – were kept on such media as the cylinder and the disk. Audio restoration techniques can do miracles in cleaning up the sound, but if essential parts of the signal were lost in the recording process, the result will sound like many historical recordings sound: flatter, monophonic. While this certainly doesn’t detract from their musical quality, the taste of the public has become increasingly shaped by digital recordings2.
Enter Zenph Studios, a company created in 2002, with a novel idea – like Welte did for recording, but now for restoration: extract from historical analogue piano recordings not the sound, but performance information: attack, dynamics, rhythms. This data can then be used to have a modern player piano reperform the recording without the scratches, hisses and clicks of the media, with all the glorious, dynamic range of the piano, and even without the humming of the pianist or the coughs of the public. The result is stunning: listen to the “reperformance” of Chopin’s Troisième prélude by Alfred Cortot, originally recorded in 1926, replayed on a concert grand in a small reverberant concert hall, and recorded on six channels. What you’ll hear is a stereo recording, but the rerecording allows for the production of immersive versions on SACD multichannel discs or on binaural spatialization systems.
The first recording which Zenph produced commercially (on the Sony label) was Glenn Gould’s 1955 interpretation of Bach’s Goldberg Variations, reperformed in the Glenn Gould Studio of Radio Canada, on a Yamaha Diskclavier Pro harmonized to sound like the 1955 instrument used by Gould.
He would certainly have liked this, his disdain for live performances being almost as legendary as his recordings. Actually, the information which has been thus extracted from his recording allows one to “correct” the playing without having to resort to additional takes: alter notes (Cortot was known to miss some), correct the attack, the intensity or the duration… anything goes. Yet at this point, all the changes will be the decision of artistic directors or sound engineers, not those of the defunct artist, and thus raises even more than before the question of truth and authenticity, while at the same time striving to achieve perfect restoration. At least it is true to the current taste. And at any rate, this process will be sure to resurrect one thing: the copyrights on these recordings, and for a very long period.
This is yet another step in the long march we have sketched. It has brought us to a day where we can go to a concert and hear the playing of the past on an instrument of the present. Add to this the technology of virtual reality, and soon we’ll be able to “see” the dead pianist perform, as if he were in the hall with us (or in a remote hall). Aren’t we on the threshold of a musical Jurassic Park? (First published in IAML Electronic Newsletter no. 27, December 2007)
1 William Leslie Sumner, The Organ. Macdonald, London, 1964.
2 Which aren’t always truer to the original signal, by the way, but in different ways. Digitization may add some artefacts to the sound, and in many cases it is “enhanced” in order to sound “better” and thus to sell better.
Voir aussi les articles (liste générée automatiquement) :