The Spin Doctor Europadisc's Weekly Column

Of Artificial Intelligence and lost music

  21st June 2022

21st June 2022


Ever keen to keep abreast of the latest trends, we’ve been given pause for thought by two recent news items and how they might relate to the world of classical music. The first was an article on the BBC News website about the impact that Artificial Intelligence (AI) is having on the world of professional tennis. Statistical analysis in the form of data analytics from matches involving leading professional tennis players has been harnessed in a range of apps and techniques. There are also video programmes that can analyse any given player’s technique. All this information and technology can then be used by coaches and players to improve their own game.

In the world of music, the closest analogy is obviously the use of recordings not for entertainment’s sake but to analyse either one’s own or another’s performance. Yet there are a number of difficulties with such an approach. Anyone who has listened to a tape recording of their own voice will know how unsettling the experience can be: what one hears from within one’s head while speaking can seem quite alarmingly different from what comes out of a playback device, however sophisticated. So it is with music: although it can be extremely useful in highlighting mistakes or (for example) awkwardnesses in phrasing and breath control, judging one’s own performance in this way can lead to paralysing self-consciousness – quite the opposite of helpful! And the dangers can even multiply when watching a video. This is one reason why so many professionals, when recording, instead use the expert pair of ears provided a record producer who, in collaboration with the technical expertise of the engineer, can enhance the strengths of a recording in ways that a self-conscious performer often couldn’t possibly match.

There are, of course, performers who have embraced technology, none more famously than conductor Herbert von Karajan: hundreds of photos of him sat by the mixing desk testify to his commanding (his detractors would say ‘controlling’) presence at all stages of the recording process. For some, his performances on disc come across as just too streamlined, too ‘managed’, compared with the more urgent spontaneity (and even occasional sloppiness) of such temperamentally different figures as Toscanini, Furtwängler, Erich Kleiber or Hans Knappertsbusch – none of them technically-minded, and to varying degrees antipathetic to the recording process. The controlling nature of Karajan’s art comes across most clearly in the videos, where ranks of brass held at precisely the same angle are the result of miming under the maestro’s beady eye to the existing audio track: an act of pure fakery whose illusion demonstrates perhaps better than anything else why Karajan and the recording industry were made for each other. At least he didn’t get orchestras to mime to their own recordings in live concerts (something not unknown in the world of rock and pop), which is probably why his live recordings – especially the really ‘live’ (i.e. single-take) ones on such labels as Orfeo d’Or and Testament – are often so much more musically interesting than the studio ones.

Another technique from the ‘pop’ world - the use of pitch-correcting ‘autotune’ programmes – has not (we hope!) been widely used in the classical world. Performers are usually trained to a sufficiently high level that, together with a sympathetic conductor and/or producer, they can normally be relied upon to deliver acceptable intonation, particularly when under the unforgiving glare of the microphone.

One area where AI is beginning to make a difference in classical music is in the realm of composition. Here, however, we are not talking about electronic composition as practised by a wide range of modern (and modernist) composers, from Varèse, Xenakis and Stockhausen up to the present day, which has an impressive pedigree at the technological vanguard. Rather, it brought to mind the recently publicised use of AI to ‘compose’ a Scherzo from Beethoven’s Tenth Symphony for which only the barest fragmentary sketches survive. Unlike an earlier scholarly completion of the same work’s first movement by the musicologist Barry Cooper, this recent job feels a rather detached affair, even when played (in a recording viewable on YouTube) by a group of clearly able and committed performers. Although the Rutgers University-based team have done an impressive job in ‘teaching’ an AI programme to compose ‘in the style of’ Beethoven, the result comes across as a piece of highly competent student pastiche, while sorely lacking the (super-)human element, the unexpected twists, the sudden emotional piercings that Beethoven of all composers was so supreme at deploying.

This delving into the realms of ‘what might have been’ brings us neatly to another recent news story (this time courtesy of The Guardian) that explored the use statistical modelling to calculate roughly how many books written in the past have been irretrievably lost to the ravages and vicissitudes of time. The related research led by Mike Kestemont and Folgert Karsdorp makes for sobering contemplation: no less than 90% of all medieval manuscripts preserving such ancient narratives as the Arthurian legends and the story of Siegfried are likely to have been lost.

How might this be reflected in the situation with music? Compared with writing, the emrgence of regulated musical script was a relatively late development, yet the possibility that hundreds, perhaps thousands of examples of medieval chant and early polyphony have been lost to events like fires, decay and artistic, religious or political censorship must be acknowledged. In England, the dissolution of the monasteries under Henry VIII will certainly have destroyed many examples of medieval and early Renaissance polyphony. It is largely because his music was so admired on the continent that John Dunstaple’s output is better preserved than that of many of his English contemporaries.

War, fire and even simple changes in fashion can account for such losses as a good fifth (a conservative estimate) of Bach’s cantata output, as well as at least a couple of Passion settings. In more recent history, Sibelius is thought to have burned the manuscript of his Eighth Symphony in a fit of self-criticism, much as Brahms, a generation earlier, destroyed all traces of a series of early string quartets that he judged not up to scratch. That is the creator’s prerogative. But unexplored libraries may yet contain unexpected surprises (dare one hope for any of Monteverdi’s lost operas?), either in as-yet uncatalogued or misfiled volumes, or even (as is often the case with earlier music) used as binding material for later, quite possibly non-musical books.

Not everything that turns up from the past will be a work of genius, and some of the discoveries announced in recent years have been less earth-shattering than the publicity bumph makes out. Yet the emergence of lost and forgotten music (as demonstrated by the re-emergence of works by such tragically fated composers as Erwin Schulhoff and Pavel Haas) can refreshen a repertoire that, without continued exploration and (re)discovery, would risk turning stale. As much as being open to newly-created music by our contemporaries, we owe it to ourselves and to music to be receptive to new discoveries when they come along, enriching and sometimes even radically changing our perspectives on the music of the past.

Further reading:
How an AI finished Beethoven’s last symphony https://www.sciencefocus.com/news/ai-beethovens-symphony/

Latest Posts


The Hard Truth about Music’s Soft Power

6th July 2022

At one extreme, there is Stravinsky’s famous dictum that ‘Music is, by its very nature, essentially powerless to express anything at all’; at the other, Aldous Huxley’s oft-repeated quote that ‘Music is what feelings sound like’. The question of whether, at an objective level, music is capable of expression is one that has preoccupied philosophers, aestheticians, composers and musicians down the centuries. It came to the fore again recently with reports of a study by academics at Western Sydney University in Australia that... read more

read more

Bringing Vaughan Williams to new audiences: An interview with Midori Komachi

28th June 2022

There’s no doubt that among this year’s musical anniversaries, the sesquicentenary of Ralph Vaughan Williams’s birth ranks high for many performers and audiences. The ever-popular Lark Ascending, the symphonies, orchestral fantasias, choral works and song settings will be well-represented, but among the more enterprising musicians exploring RVW’s musical legacy is violinist Midori Komachi. In partnership with experienced pianist Simon Callaghan, she is releasing an album containing all of the composer’s works for violin and... read more

read more

Early Music: A Stock-Taking, Part 2

15th June 2022

In the past couple of decades, Early Music can be said to have come of age. After a shaky but extremely characterful start in the early years of the 20th century, and a postwar ‘rebirth’ with the likes of Nikolaus Leonhardt and Gustav Leonhardt spearheading exciting new ventures, the 1970s and 80s can retrospectively be likened to moody teenage years, with an insistence on authenticity and dogma that raised the hackles of the traditionalist establishment. (Talk of ‘tradition’ is always problematic vis-à-vis Early Music:... read more

read more

Early Music: A Stock-Taking

8th June 2022

Our recent piece on ‘Versions and Editions’ prompted an interesting response from some of our regular readers, as a result of which we take a closer two-part look at the history and present state of the Early Music movement in the 20th and early 21st centuries.

Increasing numbers of editions of Early Music repertoires (everything from the Baroque and before) fostered a growing interest among performers. Among the first to take a deeper interest in such repertoires were instrument player and manufacturer Arnold... read more

read more

How to Celebrate a Jubilee

31st May 2022

The music industry loves anniversaries. The centenaries, sesquicentenaries, bicentenaries (and so on) of composers’ births, deaths and other significant occasions (the 2013 centenary of Stravinsky’s Le Sacre du printemps, for example) provide excuses for new recordings, reissues and box-set compilations of varying enormity and expense. Even literary figures with musical leanings (like Proust) are celebrated with themed releases, as are other significant figures from history, including royalty. So it’s a little surprising... read more

read more
View Full Archive