Re: [tuning] Digest Number 907

>I think this is probably the best birthday present I am going to get. It
>really ties together the random stuff I've learned about (neo-)gothic
>polyphony. The melody being on the bottom is really wierd, though I suppose
>there's really no reason for it to be on the top. It sounds wicked in a
>chorus voice on my synth. In Robert Walker's midi, why is one note of the
>penultimate chord in a different timbre?

Ah, found a bug in TS.

It was supposed to be choir Ahs for all the voices.

However I haven't tested combination of recording to Midi as you play
while using as Midi In the Midi output from another prog before.

Looking now at the hex dump for it, I see that it only did the Ah patch for that
one note, and the rest didn't have any voice patches at all.

Since they didn't have any patches, and because Media player doesn't do a
reset for the patches, when I played it back in Windows Media Player, it
sounded okay.

There's an option in TS to use the original patches from the MIDI In, and with
that option you get the patches okay, but if you choose option for TS to use it's
own patches, which is what I did, it leaves them out - definitely a bug.

So here is a version with choir Oohs and choir Aahs for the melody line.
http://www.robertwalker.f9.co.uk/happy_birthday_Keenan.mid

This one has Oohs and shanai:
http://www.robertwalker.f9.co.uk/Happy_Birthday_Keenan_with_shanai.mid

This one has no patches at all:
http://www.robertwalker.f9.co.uk/Happy_Birthday_Keenan_no_patches.mid

The notes are assigned to the channels in order of pitch. Not sure exactly
how that is happening at present - I think NWC is prob. playing them on
channels 1.and 2., and TS is re-assgning them to separate channels for
each voice in order to get the pitch bends right, and is ignoring the NWC
distinction between channs. 1. and 2, and making the voice distinction
between 1. and 2. by using multiple patches in same channel.

So if you play the last two in that order, you will prob. find that when you
get to the one with no patches, the shanai is playing the lowest notes only,
without the voice crossing.

(Multiple simultaneous patches in same chann. are not in Midi spec, but
I gather that they are common to the nearly all modern implementation,
e.g. of soundcards, that uses concept of tracks instead of channel, and
TS uses them extensively).

N.B. it uses channels 2 to 4, and chann 5. for the extra note in the
second last chord. (Not sure why it isn't using Chann 1 - have to look
into that...)

It should really also change channels between every note for each melody line.

That's for voices like celeste that can continue to sound noticeably after the
note has stopped (depending on Midi Synth)

When all notes are played in same channel, the resonance from the previous note
will be pitch bent along with the new note, because MIDI pitch bends are for
the entire channel rather than a single note. Can have rather wierd effect for wide
pitch bends and solo voice, but with these rich chords, you don't really notice it
when it happens.

Alternatively, you can just use an All sound off to switch off all sound in the
channel before the pitch bend, but that can create a noticeable "Click" effect
as all the voices are abruptly cut off, so I don't want to do that.

This shows what I mean (first few notes of a fractal tune, Celeste voice,
deliberate choice of gamelan scale that has wide pitch bends, including
one of -46.44 cents):
http://www.robertwalker.f9.co.uk/pitch_bent_resonances.wav
[156 Kb]

And with changes of channel for each new note that requires a new
pitch bend.
http://www.robertwalker.f9.co.uk/change_chann_for_pitch_bends.wav
[93 Kb]

For some reason TS isn't doing this change of channels in this sit.
at present, so I'll investigate and fix that, and then do a new midi clip
- may sound a bit "crisper".

This is 12 tet to compare to hear what diff. the scale makes:
http://www.robertwalker.f9.co.uk/Happy_Birthday_Keenan_et.mid
(same voicing as Oohs + shanai)

I've also uploaded the hex dumps - same file names with ext .hex
They include comments added by TS, to say what all the MIDI hex
codes do.

The cents values are the actual values in the midi file, i.e. the closest
ones you can get to within the MIDI spec for pitch bends.

N.B. for TS users:

There's also a bug in the new record to midi as you play button -
sometimes sets the patch for one channel, then plays it on another
- I had to fix it to get these midi files to do the patches properly.
Doesn't affect the Save As Midi file option.

Next upload of TS beta is delayed - trying to track down a bug that I've
introduced since last beta upload and is proving really elusive- if worst
comes to worst, may turn out to be quickest to go back to the 14th Oct code
and add in all the new things again, sorry,

Robert