Saturday, January 05, 2013

Do try this Cage-inspired listening experiment at home

‘As reported last year composers such as Hindemith, Henze, Honegger, Rihm, Penderecki, Sessions, Carter, Tippett & Milhaud continue to do poorly or are ignored altogether… Some British composers remained in the doldrums, Rubbra, Rawsthorne, Tippett, Simpson…’ - 2012 Survey of classical music broadcast on BBC Radio 3

‘…you can become open-minded, literally, by giving up your likes and dislikes and becoming interested in things’ – John Cage

‘So many riches in your photo detail! [see below] It makes one want to stay home for a week and do nothing but "blind" listen from one's own collection’ – comment by John McLaughlin Williams on the photo used in All we are saying is give chance a chance.

John McLaughlin Williams' point about “blind” listening from one’s own music collection is a very important one, but how to do it? As mentioned in an earlier post I have been experimenting with chance techniques to eliminate likes and dislikes from my home listening. The following is a description of a simple John Cage-inspired chance process to determine my home listening which has yielded stimulating results using the random number generator at random.org. Information in parentheses is an example of one chance selection from my own collection.

My CD collection is organised in alphabetical order. For first step use the substitution cipher of A=1, B=2… Z=26 - full list here. Enter range of 1 to 26 into random number generator at random.org. Selected number is cipher for first letter of selected composer’s name – if chosen letter does not correspond to composer repeat process. (Example selected random number is 20 which corresponds to T) Count number of single or CD sets of composer’s starting with that letter. (21 CDs/sets starting with T) Use that number as upper limit to select next random number. (Range for V is 1 to 21) Selected random number is chosen CD/set counting from first CD. (Chance number is 12, which is set of Tippett string quartets). If chosen CD has multiple discs (as is the case with the Tippett quartets) use number of discs as upper range limit for another random number selection (2 CDs in Tippett set, so range is 1 to 2). Selected number is disc to be played in set. (2 = quartets 4 & 5).

Variations on this chance selection process can be developed to suit CD collections organised in different ways. Compilation CDs present a problem, with Alia Vox albums presenting a particular challenge; my current workaround is to categorise by the intial letter of the conductor not composer. The random process does eliminate subjective judgement at point of selection, but clearly cannot eliminate the likes and dislikes which determine the music that makes up a personal CD collection. Extending this technique to randomly choose music from less selective online libraries such as Spotify and the Naxos music library would be a useful development (do these platforms have a random selection facility? - I am not familiar with them) while using the same chance technique to choose which internet radio station to listen to is another option. iPod’s offer a random ‘shuffle’ selection and I have experimented with this function to create radio programme playlists. But the selection is limited by the iPod capacity, and my experience also suggests the iPod algorithm is not totally random. This post is ‘work in progress’ and reader contributions are, as ever, very welcome.

Also on Facebook and Twitter. Any copyrighted material on these pages is included as "fair use", for the purpose of study, review or critical analysis only, and will be removed at the request of copyright owner(s). Report broken links, missing images and errors to - overgrownpath at hotmail dot co dot uk

11 comments:

Mike said...

I have all my music on a home server, having started ripping all my CDs over a decade ago. I can stream that* through my house with a SONOS system and randomise from the entire collection without any work on my part.


* Unfortunately Sonos' hardware index limits effectively reduce that to about 50,000 tracks so large chunks of the library have to be managed through desktop software.

Pliable said...

Mike, thanks for that and it is clear that new technology opens up many additional possibilities. I'm glad that I omitted the first step in my own chance process, which is to use the random number generator to choose between my LP and CD collection....

Love your own blog incidentally and I envy you living in that part of the world - http://mikenbondi.blogspot.co.uk/

Mike said...

Thank you!

The iPod shuffle algorithm is no longer random because people don't have a good understanding of chance/randomness, and find any minor coincidences on a small level to signify that they're not getting a good shuffle.

There are lots of articles on the phenomenon especially with respect to music shuffling.

I should add that systems like Sonos also allow to to tap into web radio, Spotify and other online streaming services as easily as your own private library. I would experiment with Spotify but France is surprisingly rather limited with respect to music availability compared to its neighbouring countries, and I don't want to subscribe to such a limited service.

Spotify has a programmable web API that inspired and suitably armed folks could use to conduct their own experiments with shuffled streaming.

Joe Shelby said...

Heh - the survey post had no mention of Takemitsu at all.

My take for my digitized collection (digging out the CDs is too much work now...just too many of them ;-) ) is that I need to go back and re-join pieces that were split by the original CD cut.

For example, Rite of Spring tends to be broken up by movements within the sides, but musically things don't break, so if you shuffle the Rite with other stuff it cuts out at very annoying moments. On some players that aren't smart enough to have 'gapless playback' (including my car), it also means annoying gaps in the middle of the work even when playing it straight.

Thus, I need to go back and re-rip the Rite (ok, ALL my versions of it) joining together all the parts of part 1 and part 2 to a more shuffle-friendly collection.

Other works I have also need to be re-joined, including Reich's Music for 18, a few other one-movement works that are split by the CD master (I think a Simpson or a few nordic composers have pressings like this).

Finally, there are some things that I can do without going into a shuffle, like It's Gonna Rain, so the shuffle algorithm needs to consider rating (which iTunes for Mac/PC does, or at least did prior to version 11 which I've heard bad things about).

Here in the states, given that most classical stations are public radio, they generally also play the 17th and early 18th century works way out of proportion to 20th century (and modern composers are utterly unheard), and some of this is not just due to audience expectation, but also that being low-funded, they have negotiated their licenses with ASCAP etc to keep their expenses low by only playing a certain (small) percentage of works that require royalty payments.

Pliable said...

Joe, thank you for remembering Toru Takemitsu. It has been in my mind for some time that I should revisit his music. So I have paused my chance experiments and am currently listening to his works for flute and guitar - http://www.prestoclassical.co.uk/r/Ondine/ODE8392

Ole Martin Halck said...

A small problem with your CD randomization procedure is that it has its own bias (apart from that of your collection): If you use it for a while, you'll find that a given composer with a rare initial will be played far more often than a given composer with a frequent initial. If you e.g. have Quantz and Xenakis as the sole Q and X in your collection, they'll each turn up in nearly 4 out of 100 selections, whereas e.g. Berg -- a composer with a small catalogue buried (I would guess) in a huge pile of music by B composers -- would stand very little chance.

In the physical case, this is easily fixed, of course, e.g. by segmenting by feet or yards of shelf space rather than by letter.

Re Joe's comment: This is a feature I've wanted from media player software for a long time: the ability to group tracks together (e.g. movements into works, or songs into albums), and shuffle the groups rather than the tracks. Maybe some have it.

Pliable said...

Ole, you are of course quite right and thanks for pointing that out.

I guess that the real solution for a CD library like mine is to scan all the discs into a database using their barcodes, and then use a random selection algorithm from that database. But that will need someone with considerably more software skills than me to write.

This thread is providing a very useful discussion about chance listening choices. But it is also scratching the surface of the whole area of music recording categorisation, cataloguing and storage.

It has long puzzled me as to why the subject of meta data about music recordings is so neglected. It is an area that interests me as I had a professional involvement in bibliographic data a while back - see link below. Yet another example of classical music not asking the right questions.

http://www.booktrade.info/index.php/showarticle/4472

Mike said...

Music metadata has been a small bugbear of mine ever since I started digitising music in the 90s.

In particular the metadata databases used by Apple's iTunes and Microsoft's Media Players are quite awful when you move out of pop/rock music to classical/jazz/world. I don't let either bit of software touch my collection, especially as you can't trust either to honour their metadata settings, and the penalty for them breaching the trust is the loss of hundreds of manhours of labelling.

It's also sad that a great many digital downloads have poor to no composer metadata. Even when browsing iTunes, Spotify etc this information is nowhere to be seen. So many times I've seen a recital disc full of etudes, nocturnes and other no clue as to the composer. The rear disc information is rarely available to browse (although so many releases of more contemporary classical music eschew any clue as to the contents - is it vocal?, chamber?, musique concrète? - the label has decided to make it a secret).

One of the online music stores I buy from said that because iTunes doesn't display composers, the publishers simply don't provide that information any more to *any* online stores.

Re Ole's comment about grouping tracks. Windows Media Player used to allow you to edit playlists to handle grouping but it removed that feature (to spite classical music fans?). Some systems allow you to simply index playlists (e.g. m3u files) to that the shuffle happens at the work level rather than the track level.

Philip Amos said...

I was hoping that someone with infinitely more technical knowledge than I might write a note re the Naxos Music Library. Thus far not, so as I do have a link to the Library I'll just say that I can see no evidence of a true random selection facility. It may be that a quasi-random selection could be made by creating a playlist and then a subset of the playlist, but it is not for a technotwit like myself to say.

It's an interesting question for there are now circa 80 000 CDs in the Library -- the catalogues, mostly complete, of around 500 labels, now including EMI. I don't know if this would be help or hindrance. There are 124 entries for Webern; 58 for Xanakis; 216 for Schumann's Piano Concerto; 61 for PMD; and 625 for Bach's Cantatas. There's an embarrassment of astonishing riches among the performances, but inevitably clunkers, so I should think that would be a factor if one could make a random selection.

Another reminder, to end, that a free link to the Library may be had at tso.ca (Toronto Symphony Orchestra) by signing up for their very random E-newsletter and then for 'Beethoven on Demand', which is indeed the NML. The free link means you can stream all of the discs in the Library, but not make playlists, as that option is open only to paying members (around $200CDN a year, I think) and students.

Pliable said...

This very productive thread has spawned a new post - http://www.overgrownpath.com/2013/01/is-classical-music-asking-right.html

Pliable said...

Nice to see someone at John Cage's publisher tweeting about this post -

https://twitter.com/socialsoprano/status/288249175559241728