DIYAPASON-L Digest #581 - Saturday, June 1, 2002
 
Re: [Residence Organs]  sampling rates (X-posted)
  by "Tony Newnham" <organist@tsnp.fsnet.co.uk>
Pipe organ available (X-posted)
  by "Tony Newnham" <organist@tsnp.fsnet.co.uk>
Re: sampling rates
  by "Ed Stauff" <ed@mewsic.com>
Re: DIYAPASON-L Digest #580 - 05/31/02
  by "Dave Milton" <dmilton@igs.net>
Re: [Residence Organs]  sampling rates (X-posted)
  by "Peter Schmuckal" <peter@schmuckal.com>
sampling rates
  by "Robert W. Taylor" <rtaylor@sockets.net>
 

(back) Subject: Re: [Residence Organs] sampling rates (X-posted) From: "Tony Newnham" <organist@tsnp.fsnet.co.uk> Date: Sat, 1 Jun 2002 09:34:39 +0100   Hi   The scanning speed for multiplexing also needs to consider another factor = - the delay time between the first note and the last - especially if you are multiplexing several manuals on the same connection. Too slow a scan = speed can cause mis-timing in chords, where notes that should sound together are slightly "smeared". This sort of problem is sometimes occurs in MIDI systems, due to trying to stuff 16 channels down one cable.   There's an article on this very problem, by Colin Pykett, available on:- www.s-c5499.fsnet.co.uk/MIDI%20for%20organists.htm (if that doesn't work knock off everything after the .co.uk and try again!   Some musicians are very sensitive to small problems like this - often = timing discrepancies that engineers would regard as insignificant.   Another point to consider is the use of pre-programmed material - I frequently produce drum rolls, etc. in MIDI files by slowing the tempo = right down, and speeding it back up for playback. This produces a repeated note that is potentially far faster than could be played manually on a = keyboard!   Hope these ramblings are of some help.   Every Blessing   Tony ----- Original Message ----- From: Christopher Sabatowich <c.sabatowich@sthedwigchurch.zzn.com> To: <PIPORG-L@listserv.albany.edu>; Residence Organ List <DIYAPASON-L@pipechat.org> Sent: Saturday, June 01, 2002 1:54 AM Subject: [Residence Organs] sampling rates (X-posted)     | Dear list, | Approximately how fast can one play? That is, what is the = approximate | duration of each note in the fastest reasonable trill? How fine a time | resolution is needed to capture the nuances of an artist? | I am currently designing a multiplexing scheme that will transmit = the | keying data via a home-brew power line modem. The multiplexers must sample | the keyboard data two to three times as fast as it could be entered. | | Christopher Sabatowich | | | | DIYAPASON-L: a Discussion list for owners and builders of their own | Residence Pipe Organs. | HOMEPAGE : http://www.diyapason.pipechat.org | List: mailto:DIYAPASON-L@pipechat.org | Administration: mailto:owner-DIYAPASON@pipechat.org | | |    
(back) Subject: Pipe organ available (X-posted) From: "Tony Newnham" <organist@tsnp.fsnet.co.uk> Date: Sat, 1 Jun 2002 11:54:59 +0100   This is a multi-part message in MIME format.   ------=3D_NextPart_000_001E_01C20963.27FE8880 Content-Type: text/plain; charset=3D"iso-8859-1" Content-Transfer-Encoding: quoted-printable   The pipe organ in Felixstowe Baptist Church (England) is being disposed = =3D of. The church would like it to go to a good home, and only require =3D nominal payment (i.e. enough to make good and redecorate the hole it =3D will leave)/   It's a 2-manual 14 stop by Bishop & Son, c.1910 vintage, tubular =3D pneumatic action to slider chests. It appears to be in excellent =3D condition, having had a clean and overhaul within the past 5 years or so = =3D (I'm trying to determine the exact date).   e-mail me privately if you want further info.   Every Blessing   Tony   ------=3D_NextPart_000_001E_01C20963.27FE8880 Content-Type: text/html; charset=3D"iso-8859-1" Content-Transfer-Encoding: quoted-printable   <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML><HEAD> <META content=3D3D"text/html; charset=3D3Diso-8859-1" =3D http-equiv=3D3DContent-Type> <META content=3D3D"MSHTML 5.00.2614.3500" name=3D3DGENERATOR> <STYLE></STYLE> </HEAD> <BODY bgColor=3D3D#ffffff> <DIV><FONT face=3D3DArial size=3D3D2>The pipe organ in Felixstowe Baptist = =3D Church=3D20 (England) is being disposed of.&nbsp; The church would like it to go to = =3D a good=3D20 home, and only require nominal payment (i.e. enough to make good and =3D redecorate=3D20 the hole it will leave)/</FONT></DIV> <DIV>&nbsp;</DIV> <DIV><FONT face=3D3DArial size=3D3D2>It's a 2-manual 14 stop by Bishop = &amp; =3D Son, c.1910=3D20 vintage, tubular pneumatic action to slider chests.&nbsp; It appears to = =3D be in=3D20 excellent condition, having had a clean and overhaul within the past 5 =3D years or=3D20 so (I'm trying to determine the exact date).</FONT></DIV> <DIV>&nbsp;</DIV> <DIV><FONT face=3D3DArial size=3D3D2>e-mail me privately if you want = further =3D   info.</FONT></DIV> <DIV>&nbsp;</DIV> <DIV><FONT face=3D3DArial size=3D3D2>Every Blessing</FONT></DIV> <DIV>&nbsp;</DIV> <DIV><FONT face=3D3DArial size=3D3D2>Tony</FONT></DIV></BODY></HTML>   ------=3D_NextPart_000_001E_01C20963.27FE8880--    
(back) Subject: Re: sampling rates From: "Ed Stauff" <ed@mewsic.com> Date: Sat, 1 Jun 2002 07:34:26 -0400   Christopher Sabatowich wrote: > Approximately how fast can one play? That is, what is the = approximate > duration of each note in the fastest reasonable trill? How fine a time > resolution is needed to capture the nuances of an artist? > I am currently designing a multiplexing scheme that will transmit = the > keying data via a home-brew power line modem. The multiplexers must = sample > the keyboard data two to three times as fast as it could be entered.   I think you're asking the wrong question. Rather, I think you should ask "what is the shortest interval (of time) that the human ear can perceive?" =   I suspect that the ear can perceive shorter intervals of time than the finger can play. I think that your answer lies in that range of frequencies where the ear stops perceiving the pulses individually and starts hearing them as a tone, somewhere around the 20hz range.   I'll be interested to hear what other folks have to say.   -- Ed   --------------------------------------------------- Edward L. Stauff, Fitchburg MA, Charlotte VT. Organist, choral director, professional substitute. Encyclopedia of Organ Stops "www.organstops.org". Software engineer, microferroequinologist, cohouser, bibliophile, lexophile, woodworker, husband, dad. "Specialization is for insects." -- Lazarus Long  
(back) Subject: Re: DIYAPASON-L Digest #580 - 05/31/02 From: "Dave Milton" <dmilton@igs.net> Date: Sat, 01 Jun 2002 07:50:06 -0400   Re keyboard sampling rates: Based on some simple tests I've done shortest duration can be around 30 to = 40 msec, but this is difficult to generate and not able to be sustained for too long. = Typically short note durations are in the 100 to 200 msec range as a minimum. As another point = of info, fast typing on a keyboard is in the range of 60 msec per character - also cannot be = sustained for long periods, and is probably not too melodic.   My keyboard scanner works { to midi } on a 16msec scan cycle { 128 points = } and seems to capture most that I can put it's way. Given above noted peak rates, a scan = cycle in the 20 msec range is probably desireable.   I think you want to keep scan time fairly short so that you can capture = the relative time difference between notes being played or released. This probably = contributes as much to the nuance of an artist as does the note duration.   In another work I read years ago, I remember that audio events that are = closer than about 110 msec are not heard as separate entities, but as single connected things, = eg echo / chorus etc type effects.   However capturing the sequence is only part of the problem - not sure what = the response time of a typical pipe valve is, let alone the time for a pipe to speak. I = suspect that average pipe valve reponse time is on order of 300 msec or slower. { I plan to do = a few measurements on some of my own stuff in a few weeks }.   Cheers, Dave Milton        
(back) Subject: Re: [Residence Organs] sampling rates (X-posted) From: "Peter Schmuckal" <peter@schmuckal.com> Date: Sat, 1 Jun 2002 12:21:22 -0700   Hello,   I'm finding this thread very interesting since I recently finished = designing and implementing a fully MIDI based relay system where MIDI signaling was used for the console to chest link.   One reason for doing this was to make the chest & pipes an independent = MIDI instrument that could be controlled w/o the console by a computer or other MIDI capable sequencer. I thought it might be interesting to be able to control pipes w/o the organization and necessary constraints that a = console imposes.   Anyway, what this means is that the console is nothing but a glorified key scanner and all the "smarts" are at the chests.   This means that only the raw keys pressed are sent down the MIDI wire from console to chest. All duplexing, coupling, borrowing, etc. is done at the chest.   A MIDI note event requires at most 3 bytes (~1ms), or 2 bytes when streaming. That means you can send a full 12 key chord (including pedal) = /w all stops and coupling on with a maximum delay of 10ms due to MIDI. There are, of course, other delays due to key scanning and other things, but I worked it out to be about 16ms max delay from key press to magnet = actuation presuming 12 truly simultaneous note events.   So with this system there is no possiblility of MIDI "choke" since only = the raw key data is sent. But just to make sure, I like to use the "2 by 4 test" by using a piece of wood the length of the manual in order to press all 61 keys simultaneously as fast as possible to see if the system chokes or shows the note smearing effect.   Of course, during normal playing when only one or two notes are changing = at any given time, the delay is more in the <5ms region.   My background is in computer speech systems and the typical feature = sampling rate for human speech is around 10-20ms. In otherwords, your ear can perceive changes in speech features which are in this speed range. It is also unlikely however skilled the organist, that they can control their articulation to within 10-20ms when pressing 12 keys.   The other nice thing about a MIDI type approach is that only the notes = that changed are sent so that the delays are deterministic and predictable, unlike the keyscan & serialize approach where there will be an = unpredicatble "jitter" in the note delay even with a single note depending on when the = key is pressed in relation to where the scan is.   Again, the main trick is not to send anything more down the hose than you need to!   Cheers,   - Peter Schmuckal   ----- Original Message ----- From: "Tony Newnham" <organist@tsnp.fsnet.co.uk> To: "Residence Organ List" <DIYAPASON-L@pipechat.org> Sent: Saturday, June 01, 2002 1:34 AM Subject: Re: [Residence Organs] sampling rates (X-posted)     > Hi > > The scanning speed for multiplexing also needs to consider another factor - > the delay time between the first note and the last - especially if you = are > multiplexing several manuals on the same connection. Too slow a scan speed > can cause mis-timing in chords, where notes that should sound together = are > slightly "smeared". This sort of problem is sometimes occurs in MIDI > systems, due to trying to stuff 16 channels down one cable. > > There's an article on this very problem, by Colin Pykett, available on:- > www.s-c5499.fsnet.co.uk/MIDI%20for%20organists.htm (if that doesn't = work > knock off everything after the .co.uk and try again! > .....    
(back) Subject: sampling rates From: "Robert W. Taylor" <rtaylor@sockets.net> Date: Sat, 1 Jun 2002 20:01:20 -0500   When Christopher Sabatowich inquired about sampling rates, I thought that an input from the past might be helpful.   The Aeolian organ music rolls in 116 note and 176 note format have step rates of 28 steps per inch. From that rate, and using the roll tempo speeds that vary from "25" to "90", one can determine the limiting resolution (time difference) of two notes being played with their start times only one step different. Roll Tempo speed is easily converted knowing that tempo "25" means the roll travels 2.5 feet per minute, at tempo "40", 4 feet per minute and so on.   At tempo "90", the roll travels 1.8 inches per second. The perforator advances the roll 28 steps per inch. This computes to about 50 possible starting events per second. Most rolls are much slower, with tempo 50 being quite common.   If memory serves me, I believe the pianocorder system scanned at 100 times per second. Based on the Organ roll resolution, 100 scans per second = would seem to be the minimum.     Bob Taylor