second wind

[calimac]

Yeah, more about my SFCV article on the Stanford historical performances symposium, because a week has passed since publication, which means it's Letters to the Editor time!

My editors asked me to keep my article well under 2000 words. Why? Because people don't like to read stuff longer than that on the web. I think that says a lot about the web, but anyway ... So there was a lot said at the symposium that got left out of the article*, and sure enough the three published letters, all apparently from non-attendees, zero in on exactly those things. So I got to write a reply (in extreme haste, owing to odd editorial customs) trying to fill in the holes a little.

The first writer was shocked to learn that musicologists are actually giving historical credence to player piano rolls, which he claims are totally unreliable. In my reply I may have shown a little testiness at his having taken an 1800-word journalistic article as a full analysis of a complex musicological problem. But his complaints all came up at the time; some were accepted and some were considered and denied. The correspondent doubts the validity of piano rolls because sometimes they differ from sound recordings by the same person. Indeed they do, but the possibility was raised at the symposium that maybe it's the sound recordings which don't accurately represent the performer's concert style.

So who knows? Certainly not I. The third writer seems to mistake me for an expert on the subject. But I'm actually just a reporter and music-lover who was there to listen and learn, and only tentatively to offer a few opinions of my own, and those more on my personal taste than on musical or historical issues.

*as those of you who asked for my full notes now know; you did all get them, yes?

Comments

[redbird]

Is it naive of me to suggest that if someone recorded both a player piano roll and a sound recording, they might both be valid and accurate even if different? A large part of why we still go to live performances is that if $musician or $band gets up on stage and plays the same pieces of music on different nights in different cities, it won't sound the same each time.

[kalimac.livejournal.com]

No, that's not naive. In fact the London computer whizzes who presented their analytical charts comparing tempo, dynamics, and regularity of notes in a few hundred recordings of Chopin mazurkas found that, for instance, three recordings of one mazurka by Arthur Rubinstein over the years differ more from each other than from some other recordings by different pianists. (The punchline that followed this was that when a recording by Joyce Hatto comes up as virtually identical to that by somebody else, it's not coincidence; and indeed that is how Hatto's recordings were proved to be plagiarized.)

There is, however, a family resemblance in all Rubinstein's playing, and you can also detect similarities between teachers and their pupils. The types of differences that the letter-writer was citing are not night-by-night variants, but fundamental differences in interpretive style. You don't expect someone to play like Jascha Heifetz one night and like Fritz Kreisler the next, not if he's being serious, not if he's making a recording.

Or do you? It was suggested at the symposium that the very act of going in and making a recording, *For Posterity* - and recording in those days was A Big Deal, not the casual act it is today - is what led to the rise of the more formal, literal playing style of later times in the first place. Piano rolls didn't have the heavy significance that sound recordings did in the 1910s and early 20s, largely because piano rolls at the time were easier to make. (It was also suggested that sound recording made performers anxious because short pieces had to be cut and/or speeded up to fit into the 4.5 minute maximum playing time of a side.)

So you can see these are deep waters. The letter-writer's mistake wasn't raising the question, but assuming that my short article had provided all the available answers.

[sturgeonslawyer.livejournal.com]

H'mmm. This implicitly answers the question I was going to ask, but I have to ask anyway: do piano rolls capture dynamics?I thought they were basically a binary thing -- if there's a hole here, depress this key -- but you imply that there are actual dynamics to the thing so that (f'rexample) if I'm playing louder with the right than the left hand the roll will play back that way. True? And, if so, how?

[kalimac.livejournal.com]

This was entirely new to me also. But apparently there were player pianos that could record dynamics in some way.

It was, however, very crude and most pianists were dissatisfied with it. The speaker on Scriabin was confident that Scriabin's airy, ethereal playing didn't sound at all like his piano rolls do when played back.

I forget which speaker said this, or which famous pianist the anecdote was about, but the story was that the pianist was approached for a recording contract by a piano roll firm that boasted its rolls could detect and play 16 distinct shades of dynamic range. "I'm sorry," replied the pianist, "but I play with 17 distinct shades of dynamic range."

[kip-w.livejournal.com]

Yes, I got the notes. Then I promptly went out of town, so I haven't been at the same computer they were on. I look forward to reading them when I've caught up with LJ and some blogs. I expect to mark some blogs as read.

Small world; my teacher in Colorado was also a pupil of Olga Samaroff (as was one of the letter writers).

QRS and Duo-Art were the two reproducing pianos I knew of (with some dynamics). One of the processes, I don't know which, apparently recorded dynamics along with the note information... years ago, I heard something about mercury switches and marks on the roll... after which, I believe, an expert roll cutter had to translate those marks to the rolls themselves. I could see Scriabin not recognizing himself after such a two-part process with human judgment intervening. Nonetheless, I love some of those piano roll recordings, suspicious absence of warts and all.

[kalimac.livejournal.com]

The player piano we had at home was QRS.

I now vaguely remember a string of intermittent holes running along one margin of the piano rolls that may have had something to do with adjusting dynamics. Also, some of the note holes, instead of being continuous, were dotted. This may have had the effect of decreasing the air pressure and consequently lowering the dynamics, but I'm not sure. I certainly don't recall the sound volume on the player piano being affected by much other than the amount of pressure applied to the activation pedals.

[kip-w.livejournal.com]

I wish we'd had a player piano at home. In the 70s, I wanted a Pianocorder, and now they have pianos that record to a computer. I can do that with what I have, but the keys don't go up and down when it plays back, so where's the fun in that? (VanBasco's Karaoke Player is a freeware program that simulates a player piano on a PC. I use it for playing with the piano rolls that Terry Smythe scans and puts up on his page.) Grandma Babbit's old piano had been a player piano, but around 1950, Dad removed the player mechanisms for reasons that were probably sound, but which seemed vastly unfair to me. The Steck grand piano in his studio at Colorado State University was formerly a Duo-Art piano, and some of the tubes could still be seen. At least I did get to play with player pianos at the music store Dad worked at for years. I'm still looking for that version of "Hindustan."

It's possible that the dynamic adjustment on your QRS piano was shut off or had lost some function over the years. Player pianos are rather temperamental. Tubes and bellows get stiff and cracky (which I expect is why Dad took the works out of Grandma's old piano), and after a while the things just hiss a lot.

Smythe, incidentally, scans a number of QRS and Duo-Art rolls along with non-expressive ones. There are even organ rolls. No orchestrions or Violinolas, though, that I know of. I think he's gotten more than 5,000 rolls available for download now in about 20 zip files.