source file: mills3.txt Date: Fri, 6 Feb 1998 00:56:15 +0100 Subject: Re: Defining pitch bend resolution From: wcooper@socrates.berkeley.edu Bill Alves made the following comment on my experiments: > Your experiments on pitch bend resolution are very interesting, but I would > like to make a suggestion. Determining pitch change by a "timbral shift" in > a rich harmonic spectrum seems to me to be a rather dicey method of > comparison, especially if you don't know what's causing it. How about > simultaneously playing a reference tone on another MIDI channel? That way > you can clearly hear a change in the beats if the pitch changes. Thanks Bill for the excellent suggestion. I repeated some of the experiments using that method with a sine wave voice, and agree it is in fact a more reliable and sensitive technique. All pitch changes that had been detected by the overtone method were confirmed as real by the reference tone technique, but in addition some new pitch differences were revealed that had escaped detection before. In general the qualitative conclusions reached earlier were confirmed. The reference tone method made it possible to tighten up the quantitative estimates for the Sound Canvas (Roland SC-55mkII). It appears that when set at a pitch bend range of +/-1, the Sound Canvas has a pitch bend resolution of 8 bits everywhere except in the pitch bend region from 8065 to 8319, where instead of the expected three pitch changes there are none at all. The beat count shows that the omission is not compensated for by a correspondingly larger pitch change across that region. This would appear to be a design flaw. If so it is an unfortunate one for microtonalists, because it introduces a gratuitous additional error of 2.34 cents into intervals whose end points are tuned by pitch bends of opposite sign (8192 being equivalent to a pitch bend of zero). If it were not for this aberration, Sound Canvas owners would enjoy a maximum error of resolution of 0.78 cents. Another unit that I tested (Fatar GS 1000R) did not exhibit this peculiar behavior. Thanks also to John Loffink for his clarifying remarks, among them: > What you have uncovered is the underlying pitch resolution or pitch > quantization of the instrument, whether adjusted by pitch bend or > any other controller, envelope, LFO, etc. Manufacturers (except > Symbolic Sound and Justonic) never publish these specs, and very few > people have investigated them. This being so we would do well to report to John anything new about the resolution of particular models that we discover by experimentation, so that he can (if willing) post the information on his useful web site (http://freeweb.pdq.net/jloffink/default.htm). --Bill Cooper, U.C. Berkeley SMTPOriginator: tuning@eartha.mills.edu From: Xou Oxno Subject: Guitars in NYC: microtonal and otherwise... PostedDate: 06-02-98 02:40:43 SendTo: CN=coul1358/OU=AT/O=EZH ReplyTo: tuning@eartha.mills.edu $MessageStorage: 0 $UpdatedBy: CN=notesrv2/OU=Server/O=EZH,CN=coul1358/OU=AT/O=EZH,CN=Manuel op de Coul/OU=AT/O=EZH RouteServers: CN=notesrv2/OU=Server/O=EZH,CN=notesrv1/OU=Server/O=EZH RouteTimes: 06-02-98 02:39:46-06-02-98 02:39:47,06-02-98 02:38:19-06-02-98 02:38:19 DeliveredDate: 06-02-98 02:38:19 Categories: $Revisions: Received: from ns.ezh.nl ([137.174.112.59]) by notesrv2.ezh.nl (Lotus SMTP MTA SMTP v4.6 (462.2 9-3-1997)) with SMTP id C12565A3.00091E09; Fri, 6 Feb 1998 02:39:35 +0100 Received: by ns.ezh.nl; (5.65v3.2/1.3/10May95) id AA09308; Fri, 6 Feb 1998 02:40:43 +0100 Date: Fri, 6 Feb 1998 02:40:43 +0100 Received: from ella.mills.edu by ns (smtpxd); id XA09299 Received: (qmail 6219 invoked from network); 5 Feb 1998 17:39:55 -0800 Received: from localhost (HELO ella.mills.edu) (127.0.0.1) by localhost with SMTP; 5 Feb 1998 17:39:55 -0800 Message-Id: <34DA9348.5B54@virtulink.com> Errors-To: madole@mills.edu Reply-To: tuning@eartha.mills.edu Originator: tuning@eartha.mills.edu Sender: tuning@eartha.mills.edu