[tuning] Sequencing [was: Another piece with "no tuning file"]

Hi John,

Some sections of the first Bach composition you tuned sound great.
But I wonder if something went awry with the program because the
piece is so long or the number of available channels was exceeded.
There are times when almost every note is very flat, maybe as much
as 60 cents. Is it possible that offset got compounded with offset at
times?

"John A. deLaubenfels" wrote:

> [Steve Sycamore:]
> >In general, I've found using any level of chorus makes a sequence poor
> >sounding and unrealistic. My preference is for a suitable level of
> >reverb which is controlled by the external effects parameter.
>
> Thanks for sharing your expertise, Steve. Can you recommend a web-
> accessible source for learning more about how the control parameters
> work in General MIDI?

There doesn't seem to be a website with a tutorial with the kind of depth
you'd like. I think a general level of education in creating excellent
sequences is really needed, but who's going to provide that? Here are
a couple of possibly useful sites:

http://www.winjammer.com/Enhancing_Title.htm
http://tremolo.harmony-central.com/HyperNews/get/MIDI.html

> >I have a sequence of Bach's Two Part Invention #13, a fairly short
> >but beautiful piece that uses most if not all 12 chromatic tones.
> >Would that do? I like it rendered by an organ, and even a thin
> >sounding organ voice works well for the piece.
>
> Yes, please e-mail it to me. BTW, I was just on your web site (looking
> to see if you'd posted this .mid there) and noticed that "Highland
> Dance" is not accessible. Having heard and liked your music, I want it
> all!
>
> JdL

I think I've fixed that and the MIDI rendition of "Highland Dance" should be
available now.

Steve

[Steve Sycamore wrote:]
>Some sections of the first Bach composition you tuned sound great.
>But I wonder if something went awry with the program because the
>piece is so long or the number of available channels was exceeded.
>There are times when almost every note is very flat, maybe as much
>as 60 cents. Is it possible that offset got compounded with offset at
>times?

A bug is always possible, and I'll be glad to check this out. Are you
referring to the b-b-bj.zip tunings? Which particular one? Could you
give me an approximate time in the piece when it sounds particularly
bad? There certainly should not be deviations of 60 cents! Even in
7-limit, a given note seldom will be outside +/- 20 cents from 12-tET,
when things are working right.

>>Thanks for sharing your expertise, Steve. Can you recommend a web-
>>accessible source for learning more about how the control parameters
>>work in General MIDI?

>There doesn't seem to be a website with a tutorial with the kind of
>depth you'd like. I think a general level of education in creating
>excellent sequences is really needed, but who's going to provide that?
>Here are a couple of possibly useful sites:

>http://www.winjammer.com/Enhancing_Title.htm
>http://tremolo.harmony-central.com/HyperNews/get/MIDI.html

Thanks! The second one is a MIDI forum, so I could always join that and
post a question.

>>BTW, I was just on your web site (looking to see if you'd posted this
>>.mid there) and noticed that "Highland Dance" is not accessible.
>>Having heard and liked your music, I want it all!

>I think I've fixed that and the MIDI rendition of "Highland Dance"
>should be available now.

Great! I'll look for it.

JdL

In off-list correspondence with Steve Sycamore, I think we've
established this his sound card is having trouble accurately playing the
tunings of b-b-b.mid and b-b-bj.mid (same sequence with chorus removed).
My techniques do add lots of pitch bends to any sequence, and I've also
experienced bizarre results on a particular card or module with some
particular sequences.

In researching Steve's observations, I made the program compile the
maximum and minimum deviations from 12-tET that a given tuning produces.
The results are, I think, quite interesting! For b-b-b, I find:

b-b-bjcs5.mid: Bend range applied: -14.0707 to 14.0739
b-b-bjes2.mid: Bend range applied: -16.7419 to 15.1923
b-b-bjcs7.mid: Bend range applied: -18.6853 to 18.7531

Notice how much more deviation the "no tuning file" version has than
the original 5-limit tuning! This is due, no doubt, to what has to
happen when two minor thirds, each targeting almost 16 cents more than
12-tET, are stacked together. The tritone thus formed targets an
interval more than 31 cents wider than 12-tET, pushing the deviations
accordingly.

I'm coming to think that, in spite of the desirable purity of the minor
thirds in the "no tuning file" versions, the original 5-limit tunings
are superior because they have less horizontal motion. If pure
intervals are the main goal, I would submit that 7-limit is the best
answer (at the obvious cost of the greatest amount of horizontal
motion).

Of course, as always, individual taste must be the main criterion for
choosing the best treatment.

JdL

[I wrote:]
>>I'm coming to think that, in spite of the desirable purity of the
>>minor thirds in the "no tuning file" versions, the original 5-limit
>>tunings are superior because they have less horizontal motion.

[Paul E wrote:]
>John, don't confuse deviations from 12-tET with horizontal motion. For
>example, a 12-tone 1/4-comma meantone tuning will have even larger
>deviations from 12-tET, but absolutely _no_ horizontal motion.

True, but in fact the "no tuning file" versions _do_ have more
horizontal motion than the original 5-limit versions:

b-b-bjcs5.dbg (2001 Feb 08 14:30:56):
7 After relaxing, Total spring pain: 185175.946
8 Final vertical spring pain: 108657.905
9 Final horizontal spring pain: 13776.655
10 Final grounding spring pain: 62741.386
[end]

b-b-bjes2.dbg (2001 Feb 08 14:32:22):
7 After relaxing, Total spring pain: 250301.414
8 Final vertical spring pain: 170704.894
9 Final horizontal spring pain: 15725.116
10 Final grounding spring pain: 63871.405
[end]

b-b-bjcs7.dbg (2001 Feb 08 14:28:44):
7 After relaxing, Total spring pain: 509237.785
8 Final vertical spring pain: 230715.027
9 Final horizontal spring pain: 52987.852
10 Final grounding spring pain: 225534.906
[end]

The es2 tuning has more pain in every category than the original cs5
does. This puzzled me till I saw the figures on max and min deviation
that I cited in my previous post. Now it makes sense.

JdL

[I wrote:]
>>True, but in fact the "no tuning file" versions _do_ have more
>>horizontal motion than the original 5-limit versions:

[Paul E:]
>Thanks for digging that up, but this paints a very different picture:
>
>cs5: 13776.655
>es2: 15725.116
>cs7: 52987.852
>
>than what you posted, which was:
>
>cs5: -14.0707 to 14.0739
>es2: -16.7419 to 15.1923
>cs7: -18.6853 to 18.7531
>
>In fact one would be inclined to say that there is no significant
>difference between the amount of horizontal motion in the two 5-limit
>versions, and much greater differences would occur if one used
>different spring strengths and/or a different piece of music.

Just for fun, let's see what wamk280 would look like, both with the
softer ('s') vertical springs and with the more rigid ('r') vertical
springs.

wamk280cs5.dbg (2001 Feb 13 14:34:18):
7 After relaxing, Total spring pain: 332110.114
8 Final vertical spring pain: 191436.499 (strength 17159.639)
9 Final horizontal spring pain: 20027.656 (strength 586398.214)
10 Final grounding spring pain: 120645.959 (strength 30631.152)
11 Bend range applied: -13.2245 to 13.4332
[end]

wamk280es2.dbg (2001 Feb 13 14:38:38):
7 After relaxing, Total spring pain: 417174.471
8 Final vertical spring pain: 285305.892 (strength 16830.183)
9 Final horizontal spring pain: 19603.189 (strength 586398.214)
10 Final grounding spring pain: 112265.390 (strength 30631.152)
11 Bend range applied: -14.5949 to 14.1860
[end]

wamk280cs7.dbg (2001 Feb 13 14:30:06):
7 After relaxing, Total spring pain: 661054.938
8 Final vertical spring pain: 334807.502 (strength 18550.221)
9 Final horizontal spring pain: 57276.634 (strength 586398.214)
10 Final grounding spring pain: 268970.802 (strength 30631.152)
11 Bend range applied: -17.0040 to 13.4705
[end]

wamk280cr5.dbg (2001 Feb 13 14:22:34):
7 After relaxing, Total spring pain: 695463.296
8 Final vertical spring pain: 366823.399 (strength 68630.456)
9 Final horizontal spring pain: 70946.069 (strength 586398.214)
10 Final grounding spring pain: 257693.827 (strength 30631.152)
11 Bend range applied: -17.1612 to 17.3104
[end]

wamk280er2.dbg (2001 Feb 13 14:26:20):
7 After relaxing, Total spring pain: 1037223.901
8 Final vertical spring pain: 680331.897 (strength 67312.631)
9 Final horizontal spring pain: 81872.663 (strength 586398.214)
10 Final grounding spring pain: 275019.341 (strength 30631.152)
11 Bend range applied: -19.8646 to 18.3975
[end]

wamk280cr7.dbg (2001 Feb 12 11:34:10):
7 After relaxing, Total spring pain: 1222826.835
8 Final vertical spring pain: 509569.280 (strength 74192.784)
9 Final horizontal spring pain: 183403.219 (strength 586398.214)
10 Final grounding spring pain: 529854.336 (strength 30631.152)
11 Bend range applied: -23.1460 to 17.2157
[end]

Wow, there's _lots_ of interesting data in there. For the softer
springs, horizontal pain _falls_ from cs5 to es2, but rises
significantly (well, by about 15%) between cr5 and er2.

Anyway, you're right, Paul, that the bend range values alone give a
picture unfair to the "no tuning file" versions. I guess what stands
out most in the data is the "price of 7". It's no wonder that, for
this reason (and perhaps others as well), people don't jump on the
7-limit bandwagon on first hearing.

JdL