3 Oct
2011
3 Oct
'11
10:20 p.m.
Don't base anything on 7-bit MIDI, it's too low-fi for a modern system, plus
it's keyboard-centric (i.e the concept of each note 'stuck' at a single
pitch is too restrictive). Plus it's being replaced by hi-def MIDI.
My suggestions:
* Use at least 32-bit floating point to represent pitch to give high
accuracy.
* Base pitches internally on octaves. E.g. 440Hz = 'octave 0.0', 880Hz =
'octave 1.0' etc. An octave is a fundamental unit of pitch which is not
biased to any particular tuning system.
Don't use plain frequency in Hertz, Hz are very un-musical because they do
not scale the same way as our ear perceives pitch.
* Separate the concept of 'key number' and 'pitch' (MIDI seems to confuse
people on this point). A key-number is NOT pitch and vice versa. This is so
you can...
* Support two 'keys' playing the same pitch, this is critical for stringed
instruments like guitars.
* Support 'bending' pitch on each note independently.
* Support the concept of re-triggering a voice that's already playing, this
is important for any percussive instrument. E.g. hitting a cymbal twice in
quick succession should not trigger the sound of two cymbals playing
together.
Best Regards,
Jeff
SynthEdit
4 Oct
4 Oct
10:16 a.m.
New subject: Internally representing pitch : A new approach
Hi,
On Tue, Oct 04, 2011 at 09:19:48AM +1300, Jeff McClintock wrote:
Don't use plain frequency in Hertz, Hz are very
un-musical because they do
not scale the same way as our ear perceives pitch.
I disagree here.
Hertz is the globally accepted SI unit to represent frequency, so IMO it
should be used in every software dealing with frequencies at the base
layer. On top of that other mappings and transform operations are
useful, which includes some kind of traditional pitch ("midi pitches")
of course, or the widely used scala format. Other "frequencies" may be
better expressed through their periods. For example 2 Hz quite often is
used as "120 bpm" or a period of 0.5 seconds. So my suggestion is: Use
Hz and (milli-)seconds internally, and provide all kinds of additional
conversions/mappings.
What is musical and what is un-musical (and even "the way our ear
works") is a very disputed issue so it should be a bit distanced from
the core of modern music software. It's maybe better to be less musical
and more scientific.
Ciao
--
Frank
11:05 a.m.
New subject: Internally representing pitch : A new approach
On Tue, Oct 04, 2011 at 10:21:53AM +0200, Frank Barknecht wrote:
On Tue, Oct 04, 2011 at 09:19:48AM +1300, Jeff
McClintock wrote:
I had the same reaction when reading 'Hz are very un-musical
because they do not scale the same way as our ear perceives
pitch'.
So what ? None of the traditional ways to represent pitch
do that. It's irrelevant.
Pitch perception depends on loudness, spectrum, the presence
of modulation (in both the musical and technical meanings of
that word actually), the presence of other signals, etc.
Do MIDI note numbers or any logarithmic scale (e.g. octaves)
capture that any better than 'Hz' ? Not at all. There are other
indications that things are not as simple as they are often
presented. For example, critical bandwidths do not obey either
a linear nor a logarithmic law.
And anyway, at some point when generating a signal and no
matter how you express pitch, you have to convert to Hz and
divide by the sample rate (also Hz).
Ciao,
--
FA
Don't use plain frequency in Hertz, Hz are
very un-musical because they do
not scale the same way as our ear perceives pitch.
I disagree here. 
6 Oct
6 Oct
5:44 a.m.
New subject: Internally representing pitch : A new approach
On Tue, 2011-10-04 at 09:19 +1300, Jeff McClintock wrote:
* Support the concept of re-triggering a voice
that's already playing, this
is important for any percussive instrument. E.g. hitting a cymbal twice in
quick succession should not trigger the sound of two cymbals playing
together.
Well, actually it should. The waves generated by the first strike are
unaware of the waves generated by the second strike and will pass
through them as if they did not exist. Compare to dropping two pebbles
in a bucket of water.
/j
7:09 a.m.
New subject: Internally representing pitch : A new approach
On Thursday 06 October 2011, at 05.43.20, Jens M Andreasen
<jens.andreasen(a)comhem.se> wrote:
On Tue, 2011-10-04 at 09:19 +1300, Jeff McClintock
wrote:
That is true for a linear system - but is a cymbal linear...?
Either way, I don't think re-trig vs new voice is all that important for that
kind of sounds. The major problem is that the human ear is annoyingly good at
finding patterns in apparent randomness, so it's probably more worthwhile to
focus on dynamics and variations, to eliminate that retro sampler feel.
However, consider a single string on a guitar. There, you *specifically* want
each new note to kill any previous note on that string, or it just won't sound
anything like a guitar. Same deal with mono synth style sounds and whatnot.
Re-trig, continous pitch and that sort of things are all about being able to
handle different kinds of sounds and play styles without cumbersome hacks and
workarounds.
Thinking about what you can do with your fingers acting directly on various
physical objects, and trying to express that information digitally might be a
good start. More "mechanical" instruments just limit that freedom a bit.
--
//David Olofson - Consultant, Developer, Artist, Open Source Advocate
.--- Games, examples, libraries, scripting, sound, music, graphics ---.
| http://consulting.olofson.net http://olofsonarcade.com |
'---------------------------------------------------------------------'
* Support the concept of re-triggering a voice
that's already playing,
this is important for any percussive instrument. E.g. hitting a cymbal
twice in quick succession should not trigger the sound of two cymbals
playing together.
Well, actually it should. The waves generated by the first strike are
unaware of the waves generated by the second strike and will pass
through them as if they did not exist. Compare to dropping two pebbles
in a bucket of water.
7:32 a.m.
New subject: Internally representing pitch : A new approach
On Thu, 2011-10-06 at 07:09 +0200, David Olofson wrote:
No it is not. Bend it enough and it will break in half :) - but within
reasonable boundaries, playing with sticks, the approximation is close.
/j
Compare to
dropping two pebbles
in a bucket of water.
That is true for a linear system - but is a cymbal linear...?
4585
days inactive
4588
days old
linux-audio-dev@lists.linuxaudio.org
5 comments
5 participants
participants (5)
-
David Olofson -
Fons Adriaensen -
Frank Barknecht -
Jeff McClintock -
Jens M Andreasen