Kameoka & Kuriyagawa

K&K are Kameoka and Kuriyagawa. Their work differs from Plomp's
and Levelts's in that they consider all possible pairs of partials and
refer to a more complex model of perception based on Stevens's concept
of a prothetic continuum. Their algorithm entails a great deal more
computation than P&L's and requires taking powers and roots with
fractional exponents. However, the results are similar to those of
the P&L algorithm -- pairs of partials within a critical bandwidth
interact and are dissonant, SPL matters, and pitches more than
an octave apart barely contribute to the overall "dissonance power."

Whether it is more accurate than P&L's isn't known, but the latter
seems to work just fine in practice.

Their principal paper is the one below; Part 1 dealt with sine waves
and the theoretical basis for their work. It immediately preceded this
one in the same volume of JASA. They also published a couple of abstracts
and some short summaries in Japanese. I have no idea what they are
doing now or whether they continued this research, which was done at
a company.

Kameoka, A. and M. Kuriyagawa. "Consonance theory Part II: Consonance
of complex Tones and its calculation method", Journal of the
Acoustical Society of America vol. 45, 1968, pp. 1460-1469.


--John



Received: from ns.ezh.nl [137.174.112.59] by vbv40.ezh.nl
with SMTP-OpenVMS via TCP/IP; Tue, 5 Nov 1996 03:10 +0100
Received: by ns.ezh.nl; (5.65v3.2/1.3/10May95) id AA00759; Tue, 5 Nov 1996 03:11:11 +0100
Received: from eartha.mills.edu by ns (smtpxd); id XA00757
Received: from by eartha.mills.edu via SMTP (940816.SGI.8.6.9/930416.SGI)
for id SAA09818; Mon, 4 Nov 1996 18:11:05 -0800
Date: Mon, 4 Nov 1996 18:11:05 -0800
Message-Id:
Errors-To: madole@ella.mills.edu
Reply-To: tuning@eartha.mills.edu
Originator: tuning@eartha.mills.edu
Sender: tuning@eartha.mills.edu

Friends,

Doesn't anyone know the details about pitch bend tuning on a K2000?
Can someone just tell me the details about using pitch bend to tune in
general? I assume that in order to do so, one must know the kind of
details which I asked about in my last post. Gary Morison, seems to agree
in his response to a different question which contained the following:

"2. As far as I can tell from the MIDI spec, there is no standard amount of
pitch deflection that a give amount of change in pitch-wheel value should
produce. I would not be surprised if there is a defacto standard, perhaps
whatever the DX-7 does or something to that effect. Similarly, I know of
no guarantee that all instruments will bend pitch consistently with each
other. By that I mean that, best I can tell from the MIDI spec alone at
least, some instruments could bend pitch in proportion to frequency, some
in proportion to log frequency, and some in proportion to period, if they
so desire. Again, there may be a defacto standard that will save you.

Perhaps somebody will correct me if I'm wrong. "

Has anyone successfully found out these sort of details for any
specific synths. I know on the K2000 at least, the response to pitch bend
messages is scalable from 1 cent to 7200 cents for full MIDI value of 8192.
But does that mean that if it is set to 1 cent that the Kurzweil has a
hardware tuning resolution of 1/1892 cents? It seems highly unlikely. So
how does one find out how specific instruments respond to pitch bend, and
how their hardware tuning is implemented? Could someone please address my
previous post, and at least tell me that no one knows if that is the case
so that I may then track down the engineers?

Any information form anyone with even the slightest relevance would
be greatly appreciated. We need to figure these details out.

Oh, and in honor of good old Pandora, does anyone know what the
current status of ZIPI is?
And has anyone used the tonality functions in Peter Stone's
Symbolic Composer?


-andrew

Safe Journey...



Received: from ns.ezh.nl [137.174.112.59] by vbv40.ezh.nl
with SMTP-OpenVMS via TCP/IP; Thu, 7 Nov 1996 00:30 +0100
Received: by ns.ezh.nl; (5.65v3.2/1.3/10May95) id AA06280; Thu, 7 Nov 1996 00:31:13 +0100
Received: from eartha.mills.edu by ns (smtpxd); id XA06272
Received: from by eartha.mills.edu via SMTP (940816.SGI.8.6.9/930416.SGI)
for id PAA15872; Wed, 6 Nov 1996 15:31:05 -0800
Date: Wed, 6 Nov 1996 15:31:05 -0800
Message-Id: <22961106232822/0005695065PK5EM@MCIMAIL.COM>
Errors-To: madole@ella.mills.edu
Reply-To: tuning@eartha.mills.edu
Originator: tuning@eartha.mills.edu
Sender: tuning@eartha.mills.edu

Oh, one other thing regarding the K2x00. I recall my K2500-user friend
concluding that it queues up control-change messages and responds to them every
so-and-so number of milliseconds. That as opposed to responding to them in
order as they come in. I don't recall what he claimed that "so-and-so" number
of milliseconds to be.

Can anybody confirm or deny that conclusion?

If indeed that is true, and if that also applies to pitch-bend messages, then
it is theoretically possible for a pitch-bend message sent before a note-on to
actually be serviced after the note-on, causing a "blurp" in the attack.


Received: from ns.ezh.nl [137.174.112.59] by vbv40.ezh.nl
with SMTP-OpenVMS via TCP/IP; Thu, 7 Nov 1996 16:09 +0100
Received: by ns.ezh.nl; (5.65v3.2/1.3/10May95) id AA06923; Thu, 7 Nov 1996 16:10:44 +0100
Received: from eartha.mills.edu by ns (smtpxd); id XA06924
Received: from by eartha.mills.edu via SMTP (940816.SGI.8.6.9/930416.SGI)
for id HAA24409; Thu, 7 Nov 1996 07:10:41 -0800
Date: Thu, 7 Nov 1996 07:10:41 -0800
Message-Id: <961107150059_71670.2576_HHB68-11@CompuServe.COM>
Errors-To: madole@ella.mills.edu
Reply-To: tuning@eartha.mills.edu
Originator: tuning@eartha.mills.edu
Sender: tuning@eartha.mills.edu

Friends,


I am attempting to use pitch bend messages sent by sequencer to a
K2000 to set frequency output to conform to JI ratios in a more perfect
fashion than possible using Global software tuning or other cent based
methods.

I sent the following message to the list last week:

> This is a rather basic question I know, but does anyone know what
>the tuning resolution of the K2000? How many steps are possible between a
>semitone, and if this number is something like 2048 or hopefully 4096, how
>are translations made between this number and cent values? Thanks in
>advance.

_______________________________________________________________________________

I recieved the following response from John at Kurweil:

If you're speaking in terms of the intonation tables, resolution is
limited to 1 cent steps by software. The value set in software is
truncated, not rounded, to the next lowest hardware resolution, which is on
the order of
1/20 cent. Transposing samples downwards 4-5 octaves or more reduces the
tuning resolution.

Changing pitch by other means such was the pitch wheel will usually
be more limited by the controller than the K2000 internals. Most if not
all of the
pitch wheels of current synthesizers have a resolution of only 7-8 bits, even
though the MIDI parameter has a resolution to 14 bits.

_______________________________________________________________________________


I still have several unanswered questions concerning generally
using the pitch benf message to control tuning, and more specifically doing
this on a K2000. Let me further clarify my query with the following
questions:

1) Where does the 1/20 of a cent figure come from? Does that mean
that there are 2000 (or probably 2048) possible frequency divisions between
a musical half step?

2) In the hypothetical scenario where the pitch bend range is set
to 100 cents under the "Common" page, and MIDI pitch bend messages are sent
using an external sequencer to produce every possible value in step size of
1 unit (0,1,2,3... ...8189,8190,8191) from 0 to 8191, how many unique
pitch values will the K2000 produce? Is it capable of producing 8192
unique values? Is the figure dependent upon the initial frequency of the
note?

3) Would I be correct in assuming that pitch bend messages affect
the absolute output frequency in an exponential fashion as in the 12Tet
model? What I mean is, if I played A440 and with a pitch bend value of 256
(and pitch bend range was set to 100 cents) would the output frequency in
Hertz be equal to 440((2^(1/12))^(256/8192))=440.7949Hz?


thanks...

-andrew

Safe Journey...



Received: from ns.ezh.nl [137.174.112.59] by vbv40.ezh.nl
with SMTP-OpenVMS via TCP/IP; Fri, 15 Nov 1996 11:02 +0100
Received: by ns.ezh.nl; (5.65v3.2/1.3/10May95) id AA11462; Fri, 15 Nov 1996 11:04:08 +0100
Received: from eartha.mills.edu by ns (smtpxd); id XA11479
Received: from by eartha.mills.edu via SMTP (940816.SGI.8.6.9/930416.SGI)
for id CAA18012; Fri, 15 Nov 1996 02:03:44 -0800
Date: Fri, 15 Nov 1996 02:03:44 -0800
Message-Id: <961115100106_71670.2576_HHB42-3@CompuServe.COM>
Errors-To: madole@ella.mills.edu
Reply-To: tuning@eartha.mills.edu
Originator: tuning@eartha.mills.edu
Sender: tuning@eartha.mills.edu