A Demonstration of How the Aeolian Duo-Art Expression System Works

[@phooesnax]

Best explanation on KZbin. Did my upright 20 years ago and had to do research for a year before starting. Took quite a while to sort out the working strategy. Properly set up these will give chills. Wonderful system. My opinion is since there are unlimited number notes played there are far more than 16 subtle levels. Bravo. Nice job,

[@KawhackitaRag]

Yes, there are numerous articles in back-issues of the AMICA "Bulletin" magazine (journal of the Automatic Musical Instrument Collectors' Association, the main American player piano society), explaining how the various interactions of the Duo-Art devices result in actually more than 16 (or even 16 times 16, if you add the theme levels to the accompaniment levels) levels, and how the more astute roll arrangers at the company were able to take advantage of this in their roll coding. This is good, since it helped the Duo-Art compete with other systems with no fixed dynamic levels, which work via slow and fast crescendos / decrescendos, those having theoretically 'infinite' levels in between the loudest and softest.

[@victoriaparkflyer848]

I just purchased an upright 1920's George Steck with a Duo Art Reproducing Player. I own a few normal player pianos . This video is outstanding for understanding how the Duo Art reproducing system works and why pianos equipped with this system are very special. I sincerely appreciate the time you took to create the video and posting it! Thanks you!

[@leilanirocks]

Very detailed explanation! Thanks.

[@evil-wombat]

Love the detailed explanation. That's essentially a four-bit binary encoding, implemented using accordion segments that successively double in size. Very clever!

[@randelanderson6754]

What a great demonstration! I have a Dou-Art and this gives me a much better understanding on how the expression system works.

[@phooesnax]

Great video! When I did our Stroud 20 years ago I spent a year investigating how this all worked before starting work. It is a tremendous system and despite 16 levels the actual levels are infinite due to choice of number of notes played at once and the interplay of the regulators on the expression box.

[@organophone]

Gongratulations for this video that helped us to restore our duoart, the expressions are perfectly set, for our duoart it is a model 1917 and it does not have a crash valve and the system is differently built, it has for pianissimos a felt muffler

[@mikemount4411]

Thanks much for this. I am setting up a Steinway XR at this time and found this very educational.Mike W. Mount

[@PiotrBarcz]

Wow! I never knew the Duo-Art was so simple!

[@KawhackitaRag]

It's simpler than some other reproducing systems. The Red and Licensee type Welte (the first/original reproducing piano from 1904) are also simple but they're 'floating' systems with no real fixed dynamic level (although they have a gadget called a 'mezzo-forte hook' that can limit the travel of the floating expression device to either the lower dynamic levels or the higher dynamic levels, keeping it from getting any lower or higher). The Recordo and Art-Apollo systems are even simpler, although they're rare and very rare, respectively. The Recordo is more like the Duo-Art with the binary / additive system of dynamic coding. Other systems (Ampico, Art Echo, various Hupfeld systems etc) have a mix of additive and floating expression devices. The additive systems have the advantage of responding well at various roll speeds / tempos (within the parameters that the system can actually work, of course), whereas the 'floating' systems or crescendo-type systems need fixed or nearly-fixed roll tempos, so that the dynamics coded into the roll 'hit' at the right time in the music (if you play a Welte roll too fast or slow, the dynamics will happen on the wrong notes!).

[@PiotrBarcz]

@@KawhackitaRag Sounds like Welte had some pretty bad drawbacks...

[@brentcoombs5293]

Makes me wonder why they're quite rare compared to "normal" Player-Pianos? Were they THAT much more expensive to make, and/or was it because of the vagaries of getting enough "famous" Piano Players on board? Thanks for posting.

[@PiotrBarcz]

@@brentcoombs5293 They had several thousand more parts, the rolls were harder manufacture, the coding and arranging was more intricate and many famous composers played their pieces on the rolls. That and also that grand pianos were more commonly equipped and as we both know, grand pianos were extremely expensive then and they're still expensive now

[@brentcoombs5293]

@@PiotrBarcz "several thousand more parts" - wow! No wonder then that My Stroud Duo-Art was so heavy to get on and off my trailer. Now if only I can find enough affordable Audiogragh/Duo-Art Rolls to justify that effort. Cheers.

[@lionelrisler2657]

Très intéressant

[@martinadler73]

Most interesting! Many thanks.

[@marcgoodman4228]

Excellent. Good and rare to hear a well-adjusted reproducing piano. So many sound like a washing machine! Would be good to give some credit to the pianist. I think it is Arthur de Greef?

[@chem100]

The most musically expressive of all reproducing piano mechanisms.

[@ozzietadziu]

Not for "Ampico" people!

[@KawhackitaRag]

The Duo-Art, Welte (various types), Ampico (all types), Art-Echo, Artrio-Angelus etc all have very good expression if properly restored and well-regulated, keeping in mind the size of the piano, size of the room etc (since some small pianos don't sound so good when 'banged' fortissimo, so the expression adjustments for those reproducing pianos need to be very carefully done so the system plays it more like a real pianist would play it who can actually hear the sound of the piano).

[@ozzietadziu]

A wonderful treatise except that your description of reproducers leaves out the "Expression systems" such as the Recordo.

[@jbum]

The expression mechanism seems really straightforward to me, as compared to the Ampico. In particular, the use of powers of 2 (1/2, 1/4, 1/8, etc.) is a power-of-2 system that maximizes the use of different selectors, vs the Ampico's 2/4/6, which provides only 7 unique combinations from 3 holes (instead of the 8 that power-of-2 ratios would provide). Made me wonder if ampico was avoiding violating a patent...

[@andrewbarrett1537]

Jim Bumgardner Well, the Stoddard Ampico was introduced in 1912, two years before the Duo-Art.

[@kenreighard8493]

The power-of-2 system also mimics the way in which the human ear responds to sound pressure level. Our ear/brain system has a logarithmic response to loudness which spans over ten orders of magnitude from threshold of hearing to threshold of pain. Either one of Aeolian Duo-Art's design engineers was very current in his knowledge of theory (the response of human hearing was a research subject at Bell Telephone Laboratories in the early 1900's,) or they did a good bit of experimentation to arrive at that system empirically. The ear's response to pitch is also logarithmic; the frequency in Hertz doubles for each octave up the scale. In an equal-tempered tuning ("stretch" is being overlooked in this example,) a semitone is a frequency ratio of the twelfth root of two.

[@andrewbarrett1537]

It's also possible that Stoddard was thinking in terms of written music DYNAMIC MARKINGS where most music is only notated at about six levels of NOTATED dynamics: pianissimo (pp); piano (p); mezzo-piano (mp); mezzo-forte (mf); forte (f); and fortissimo (ff). Of course there are also rarely extreme dynamics notated: ppp and fff for extremely soft and loud respectively. I presume the seventh level intended was a true mezzo level between mezzo-piano (medium soft) and mezzo-forte (medium loud). So, given that early Ampico coding was apparently manually created and added by the roll arrangers in post-production (after a person in the studio thoroughly marked up a score in real-time with dynamics while listening to the pianist play), rather than live-recorded, then I presume the seven-level system was fairly easy to correlate with written "mp", "ff", etc. etc. I mean EARLY and MIDDLE PERIOD Ampico dynamic coding; in the late 1920s, Clarence Hickman developed the "spark chronograph" recording apparatus, which could really record dynamics in real-time and was state of the art for the time. However, the depression hit quickly after this so few LATE commercially-issued Ampico rolls were ever actually recorded using this equipment. Most of the rest (at least popular, if not classical rolls) were simply hand-played using the old method or else totally arranged (such as those by Frank Milne). You can read more about the 'spark chronograph' here: www.mmdigest.com/Tech/Ampico/jasa.html

[@Folboi]

What do the controls say??

[@Graters1989]

what roll is that?

[@PiotrBarcz]

Do the snake bite holes do accents?

[@antoinebihr8921]

Thank you very much for taking the time to explain technically all the subtile variations, I had aways winedred how the duo-arts systems worked. I had a question, the pression controls the force, but do you know if the "attack" (the speed of the key) is controlled by any device, or is it linked with the rpessure too ? Thanks again !

[@bartolomochristofari]

+Antoine Bihr I'm glad you found the video to be useful. "Attack", key velocity, hammer velocity, hammer energy, sound energy, note loudness, etc. are all slightly different ways of describing the same musical effect. In a pneumatic player piano it is completely linked with air pressure, just like "attack" is linked with finger pressure when the piano is played by hand. More pressure makes the key move faster, putting more energy into the hammer, making a louder sound, with greater "attack."

[@KawhackitaRag]

Yes it has to do with basic physics... how hard the hammer hits is dependent upon its velocity (speed) at the time it hits the string, which then translates into force and sonority/tone. Of course, the piano stack suction levels create the initial driving force of the back of the key (and then thru the piano action to the hammer), so it is force translated into velocity which then is translated back into force as soon as the hammer hits. In fact, the very late experimental Ampico "spark chronograph" real-time dynamic-recording device (introduced right as the depression hit in 1929, so it apparently wasn't used very much), worked exactly by measuring the speed of the piano hammer from the beginning to the end of its stroke as compared with more 'absolute time' as meted out by a stopwatch type apparatus, I believe. This was state of the art for 1929 of course.

[@KawhackitaRag]

One thing that reproducing pianos CAN'T do, that live pianists can do, is lift their stack 'fingers' further off the (backs of the) keys (or wippens in an upright), in order to attack louder notes with more 'wind up' velocity, kind of like a baseball batter swinging at a ball. However, since the piano key has a limited travel (set by the piano technician), between two more or less fixed points, it can only move a short distance, when the pianists fingers finally 'hit' and start pushing it, from however high they've dropped. So the theory was that so long as the stack suction levels could mimic and reproduce the travel of the key from when the fingers hit to the end of the travel, that was good enough, and our ears tell us that it is.

[@PiotrBarcz]

So I was right! The theme holes accent the notes by a tiny step!

[@PiotrBarcz]

I'm just curious, do you know where I could possibly buy a push-up player?

[@KawhackitaRag]

You can find them for sale on the group "Player Pianos Free, Cheap and Sometimes Not... From Any Source!" on Facebook. You can also find them on the much older site "Player Piano and Mechanical Music Exchange" elsewhere on the web. If you join AMICA and post a want ad in the AMICA Bulletin magazine, you will be able to find many for sale. You can also post a want ad in the MMD (Mechanical Music Digest) newsgroup.

[@PiotrBarcz]

@@KawhackitaRag Thanks!

[@bryangl1]

I must congratulate you on the quality of restoration and reglation of th piano. An excellent demo of a reproducing piano. But the Duo-Art did not cover all 88 notes, but only 80! - a rather basic error. And while your description is good as far as it goes, more recent investigations of the expression action (particularly by an Australian authority on reproducing pianos who has just gained a PhD on the subject) reveal much greater subtleties than described here. However, this is a good, careful and well-presented introduction to what is proving to be significantly more complcated than given here.

[@bartolomochristofari]

Your comment that the Duo-Art system only played 80 notes rather than all 88, was true for later designs. Steinway (and most other makers) switched to 80-note stacks about 1926, eliminating the need for complicated cut-out blocks on earlier designs like the 1922 instrument demonstrated here that allow it to play all 88 notes when the Duo-Art system is switched off. When first introduced, Aeolian apparently believe that many customers would frequently play non-Duo-Art rolls and care about all 88 notes working, but by the late 1920s they realized that most owners played mostly DA rolls. The shorter 80-note stack allowed cheaper manufacturing, looked better with single-leg furniture styles, and eliminated capability most customers didn't care about very much. As with all complex systems, there are subtleties I may not address in a 15-minute video--but much that has been written about the system by "experts" is simply not true. Many erroneously believe that the DA knife valve is a "airflow regulator" rather than a "pressure regulator", and is somehow fundamentally different from Ampico or Welte regulators in that regard. All air-pressure regulators like this will have different response characteristics to how quickly pressure changes occur and how well pressure is maintained when airflow is high, i.e., when playing many notes at once. And the DA coding probably took this into account, tweaking the coding for typical response speed, and for slightly higher loudness levels than the artist actually played for larger chords, etc. But overall, this demo shows a true and fairly complete representation of the core system operation.

[@bryangl1]

I'm sorry, you really show a significant flaw in your knowledge. Duo-Art were ALWAYS 80-note. An ablity to switch off the Duo-Art sysem to play 88-note player rolls does NOT make the instrument 88-note DA, which is what you are implying. All Duo-Art rolls are 80-note, and a friend has a 1914 Duo-Art Steinway; 80-note as ever. You must be confusing with Themodist or some such as the Duo-Art was essentially a progression from Themodist (albeit a substantial one). Duo-Art were proud that although over time changes were made to the reproducing system, all Duo-Art rolls/pianos were compatible (not a 100% accurate statement by them). If so, it's obvious an "88-note Duo-Art" roll would not be compatible - if it existed. I realise the subject of expression control is indeed complex and would take a great many hours to cover properly. And within 15 minutes you have done it well. except you give no clear indication that it is a much more complex issue, the details of which are still being unravelled (possibly because you are unaware of current research in this area). Current researches being undertaken are accurately measuring pressures etc with precise digital devices (rather than a water gauge) and are revealing much greater subtleties and complexities in expression (which the Duo-Art engineers/musical experts must have sorted out by their musical acuity as such measurements were impossible then - they really new their stuff) than were previously realised. I'm not the expert on this, but it is the subject of a forthcoming research paper by a person who IS an authority - not a so-called 'expert'.

[@bartolomochristofari]

Nowhere in the video do I state or imply that DA-coded rolls play all 88 notes of the standard modern piano compass. Yes, I know well that coded rolls do not play the top 4 or bottom 4 notes. But your insistence that I somehow stated they did, or that this is a basic flaw in my knowledge is not correct. Certainly the DA instrument in this demonstration has a full 88-note stack. Yes, most original rolls codings were carefully edited until they sounded "good" and "right" in playback, so some subtle effects may be there, including that some notes might possibly be played at "intermediate levels", for example at level 4 1/2 rrather than just at level 4 or level 5, with careful dynamic timing of accordion movement. But your idea that people until now can't sufficiently understand significant details of DA expression because they have lacked sufficiently precise pressure gauges is verifiably not true -- whether or not such rarified knowledge comes from someone who, in your words, "IS and authority". Ultra-precise measurements might tell you a litte more about exactly how a particular Duo-Art piano plays a particular roll. But no one piano design is the "gold standard" for how the DA system should behave, and standard gauges are more than precise enough to measure significant variations found in in different DA systems in different pianos manufactured with different designs at different times. Grand vs. upright, grands of different length with different keystick balancing, variations in hammer mass and hammer voicing, slightly different regulator springs, cloth stiffness and spring adjustment on accordion pneumatics, leak rates of different stacks, valve facings and pneumatic fabrics, slight variation in roll perforation, air motor calibration, tubing run length, cross-valve vs. round valve, etc., etc. All of those things create slight but measurable variability in playback and musical result on different instruments. So it is a fools errand to believe that ultra-precise pneumatic measurements on any single instrument will provide hitherto unknown insight into important overall system attributes. Oh, and by the way, I also have a Ph.D. from a school that might be considered the top engineering and science institution in the US. I won't use that credential to claim to be "the authority" on the DA system or anything else. But the DA system is fundamentally simple--exactly 10 bits of digital expression control on top of the normal note and pedaling information in a standard non-reproducing piano roll. Claims that it somehow is complex beyond the comprehension of us mere mortals because some "authority" has weighed in are simply not true.

[@bryangl1]

Briefly: you may not have explicitly stated that early DA rolls were 88-note, but that is the implication. Listen to your opening sentences. Talking specifically in the context of the Duo-Art piano, you discuss it playing 88 notes. That is a clear (but now, apparently unintentional) implication that it is an 88-note DA action. I didn't say that the DA system was complex beyond comprehension, only that it is being found to be more complex than you are aware. I'm sure if you were familiar with recent research you would understand what is being found. The "authority" I refer to gained his PhD recently (a joint effort between Stanford U and University of Sydney) specifically on the detailed study of the main reproducing systems and what was "going on" as it were in the actions. [He also lectures around the world on the subject.] The work leading to his PhD revealed the DA was more complex than it appears and is leading to further research. The PhD was the culmination of many years of study and research particularly of the DA - except it turned out that there's room for further study. Without itemizing your reply in detail, there is much there I would not dispute (ie., variations from instrument to instrument), but as a qualified engineer, to imply that a water guage is going to reveal all that is needed to investigate subtle pressure changes and factors that influence rates of change rather than modern measuring devices, and to say, or imply, that the DA system can't possibly have any more to it than you are suggesting is revealing a rather closed mind. On which basis, further discussion with you is pointless. Stay happy in your restricted world.

[@ChuckMcC]

Wonder if it played demonic music if you loaded the paper in backwards?lol

[@PiotrBarcz]

I actually have loaded the roll in backwards in my piano. The song was flipped and was in minor! The only problem is, you can't re-wind because the slotted end of the roll isn't locking with the slotted chuck...

[@andrewbarrett1537]

I am pretty sure that most reproducing pianos would not play correctly backwards, as the expression tracks are timed to SLIGHTLY ANTICIPATE notes they are supposed to make softer or louder. So the dynamics would be completely wrong in addition to the music being backwards. But I guess you could have fun with it?