6 Oct
2013
6 Oct
'13
10:59 a.m.
A friend asked me if I had a Theramin sound. I figured, I could cheat with a sine wave or
a deeply-filtered sawtooth maybe, but then I thought, hmm, what is a real Theramin wave
anyway?
Google Images pumped me up nicely:
http://www.horst-theremin.com/images/_mid%20600%20Hz.jpg
Diagrams for waveforms of the original Lev Termen and RCA instruments, and the Moog and
Big Briar Theramins, that I was able to find, also look similar to the above.
And, umm, that ain't no sine wave. It's got some strange bias, louder on the
positive than the negative side, more rounded and sine-like on the negative side, and more
like a sawtooth on the positive side. I'm told it's also non-linear and the
waveform for the lower and higher notes look slightly different, but one thing at a time.
So I thought, well, maybe I could hack up a LADSPA oscillator plugin that does that weird
waveform.
Looked at the SDK, and sure enough, in the sine.cpp file, there's an
initialize_sine_table() that looks like the right place to put the wave. It's just a
table of floats, looks like.
However, my DSP and maths skills stop right about there. I have no idea how to generate
the points on that oddball waveform in such a way that they could be dumped into, say,
g_pfSineTable.
Has anyone already done this? Anyone want to take a shot at it?
-ken
6 Oct
6 Oct
12:55 p.m.
On Sun, Oct 06, 2013 at 01:59:15AM -0700, Ken Restivo wrote:
You don't need to reproduce the waveform, but the spectrum,
which is roughly:
Freq Amp
---------------
1 1.000
2 0.191
3 0.065
4 0.041
5 0.019
6 0.012
Apart from that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
1:25 p.m.
On 10/06/2013 06:55 AM, Fons Adriaensen wrote:
On Sun, Oct 06, 2013 at 01:59:15AM -0700, Ken Restivo
wrote:
You don't need to reproduce the waveform, but the spectrum,
which is roughly:
Freq Amp
---------------
1 1.000
2 0.191
3 0.065
4 0.041
5 0.019
6 0.012
Apart from that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
Ciao,
Fascinating.
http://www.ted.com/talks/pamelia_kurstin_plays_the_theremin.html
10:29 p.m.
On Sun, Oct 06, 2013 at 10:55:15AM +0000, Fons Adriaensen wrote:
On Sun, Oct 06, 2013 at 01:59:15AM -0700, Ken Restivo
wrote:
You don't need to reproduce the waveform, but the spectrum,
which is roughly:
Freq Amp
---------------
1 1.000
2 0.191
3 0.065
4 0.041
5 0.019
6 0.012
Cool, thanks.
How would one do that with LADSPA plugins? 6 sine wave oscillators into a mixer, one for
each harmonic?
I vaguely remember seeing an oscillator (maybe in AMS?) that had mixer settings for
various harmonics, but I'm pretty sure it only went up to the first 3.
Also, were you really able to come up with that harmonic recipe just by looking at that
waveform? If so, I find that remarkably impressive.
Apart from that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
True, that has a lot to do with the sound. I was thinking a ribbon controller or a simple
X-Y trackpad (or an OSC thing on an Android tablet) would take care of that.
-ken
10:38 p.m.
On Sun, 6 Oct 2013 13:29:23 -0700
Ken Restivo <ken(a)restivo.org> wrote:
To get somewhere near this I put a light delayed vibrato on it, enabled
portamento and used the pitch wheel on my keyboard. A touch of reverb and I was
nearly doing the effects of 'The Forbidden Planet' :)
--
Will J Godfrey
http://www.musically.me.uk
Say you have a poem and I have a tune.
Exchange them and we can both have a poem, a tune, and a song.
On Sun, Oct 06, 2013 at 10:55:15AM +0000, Fons
Adriaensen wrote:
I had a go at this with AdSynth in yoshimi but the harmonic amplitude only
has 64 steps (which I think are linear). I worked to the nearest approximation,
which left 6th harmonic at zero. It sounds sort-of right but not totally
convincing.
On Sun, Oct 06, 2013 at 01:59:15AM -0700, Ken
Restivo wrote:
You don't need to reproduce the waveform, but the spectrum,
which is roughly:
Freq Amp
---------------
1 1.000
2 0.191
3 0.065
4 0.041
5 0.019
6 0.012
Cool, thanks.
How would one do that with LADSPA plugins? 6 sine wave oscillators into a mixer, one for
each harmonic?
I vaguely remember seeing an oscillator (maybe in AMS?) that had mixer settings for
various harmonics, but I'm pretty sure it only went up to the first 3.
Also, were you really able to come up with that harmonic recipe just by looking at that
waveform? If so, I find that remarkably impressive.
Apart from that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
True, that has a lot to do with the sound. I was thinking a ribbon controller or a simple
X-Y trackpad (or an OSC thing on an Android tablet) would take care of that.
10:59 p.m.
On Sun, Oct 06, 2013 at 01:29:23PM -0700, Ken Restivo wrote:
It would have to have sufficient bandwidth and resolution,
both very high on the original.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
How would one do that with LADSPA plugins? 6 sine
wave oscillators into a mixer, one for each harmonic?
Yes, but 3 or 4 would probably be enough. Anything below
0.01 is suspect anyway.
I vaguely remember seeing an oscillator (maybe in
AMS?)
that had mixer settings for various harmonics, but I'm
pretty sure it only went up to the first 3.
No, it goes up to 20 or so. But you'll have zillions
of envelope parameters to adjust. Or just use three
simple sine oscillators.
Also, were you really able to come up with that
harmonic recipe just by looking at that waveform?
If so, I find that remarkably impressive.
Of course not :-) I digitised the waveform using a
simple program that writes mouse clicks to a text
file. Then a few lines of Python:
#!/usr/bin/python
from math import *
X = [ 0.00, 0.70, 1.30, 2.00, 2.70, 3.25, 4.00, 4.65, 5.15, 5.85,
6.35, 6.95, 7.40, 7.95, 8.30, 8.75, 9.10, 9.55, 10.00, 10.25,
10.55, 10.90, 11.20, 11.53, 11.70, 11.40, 10.98, 10.10, 9.10, 8.00,
6.90, 5.75, 4.45, 3.20, 2.10, 1.01, 0.00, -1.00, -1.98, -2.90,
-3.70, -4.25, -5.00, -5.60, -6.10, -6.60, -7.02, -7.35, -7.80, -8.00,
-8.15, -8.30, -8.40, -8.46, -8.50, -8.50, -8.47, -8.45, -8.27, -8.10,
-7.99, -7.80, -7.55, -7.25, -7.00, -6.67, -6.35, -6.05, -5.65, -5.30,
-4.98, -4.50, -4.05, -3.60, -3.05, -2.60, -2.00, -1.45, -0.80 ]
N = len(X)
def analyse(f):
sx = 0
sy = 0
w = f * 2 * pi / N
for i in range(N):
sx += X [i] * sin (i * w)
sy += X [i] * cos (i * w)
m = hypot (sx, sy)
a = atan2(sy, sx) * 180 / pi
return m, a
for i in range (1, 11):
m, a = analyse (i)
if i == 1: m1 = m
print ("%3d %10.3f %8.1f" % (i, m / m1, a))
Apart from
that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
True, that has a lot to do with the sound. I was thinking
a ribbon controller or a simple X-Y trackpad (or an OSC
thing on an Android tablet) would take care of that. 
11:12 p.m.
On Sun, Oct 6, 2013 at 9:59 PM, Fons Adriaensen <fons(a)linuxaudio.org> wrote:
#!/usr/bin/python
from math import *
X = [ 0.00, 0.70, 1.30, 2.00, 2.70, 3.25, 4.00, 4.65, 5.15, 5.85,
6.35, 6.95, 7.40, 7.95, 8.30, 8.75, 9.10, 9.55, 10.00, 10.25,
10.55, 10.90, 11.20, 11.53, 11.70, 11.40, 10.98, 10.10, 9.10, 8.00,
6.90, 5.75, 4.45, 3.20, 2.10, 1.01, 0.00, -1.00, -1.98, -2.90,
-3.70, -4.25, -5.00, -5.60, -6.10, -6.60, -7.02, -7.35, -7.80, -8.00,
-8.15, -8.30, -8.40, -8.46, -8.50, -8.50, -8.47, -8.45, -8.27, -8.10,
-7.99, -7.80, -7.55, -7.25, -7.00, -6.67, -6.35, -6.05, -5.65, -5.30,
-4.98, -4.50, -4.05, -3.60, -3.05, -2.60, -2.00, -1.45, -0.80 ]
N = len(X)
def analyse(f):
sx = 0
sy = 0
w = f * 2 * pi / N
for i in range(N):
sx += X [i] * sin (i * w)
sy += X [i] * cos (i * w)
m = hypot (sx, sy)
a = atan2(sy, sx) * 180 / pi
return m, a
for i in range (1, 11):
m, a = analyse (i)
if i == 1: m1 = m
print ("%3d %10.3f %8.1f" % (i, m / m1, a))
Wauw, that's a golden nugget of Python! Cheers, -Harry
7 Oct
7 Oct
9:38 p.m.
On Sun, Oct 06, 2013 at 08:59:21PM +0000, Fons Adriaensen wrote:
On Sun, Oct 06, 2013 at 01:29:23PM -0700, Ken Restivo
wrote:
It would have to have sufficient bandwidth and resolution,
both very high on the original.
Ciao,
I once emulated the frequency modulation with an arduino and a cheap IR sensor.
I don't remember the resolution of the arduino/IR sensor, but I got pretty good
results.
I translated the distance values into frequency values, and I remember I had to
stabelize the input, by taking avarages over the values. If you used two sensors
like this, I think you could make a nice theremin controller...
Greetings,
lieven
Apart
from that the sound will be frequency and amplitude
modulated by minute movements of the player's hands.
True, that has a lot to do with the sound. I was thinking
a ribbon controller or a simple X-Y trackpad (or an OSC
thing on an Android tablet) would take care of that. 
8 Oct
8 Oct
4:50 p.m.
On 06/10/2013 22:29, Ken Restivo wrote:
On Sun, Oct 06, 2013 at 10:55:15AM +0000, Fons
Adriaensen wrote:
You might look into using Pure Data: you can both have:
- (virtually) any number of oscillators with the advantage that each
partial's strength (and frequency) can be independently and dynamically
modified
- the [tabosc4~] object which is a wavetable oscillator. The [sinesum(
message sent to a table (a.k.a. array) with partial strengths will do
the sum for additive synthesis. In this case of course it woun't be
possible to dynamically change the partials' strength (although I think
people have used tricks to switch between different tables seamlessly).
Not to mention the many built-in features to connect to the external
world for control such as MIDI, OSC, TCP/IP, arduino, to mention a few..
Hope this helps.
Lorenzo
On Sun, Oct 06, 2013 at 01:59:15AM -0700, Ken
Restivo wrote:
You don't need to reproduce the waveform, but the spectrum,
which is roughly:
Freq Amp
---------------
1 1.000
2 0.191
3 0.065
4 0.041
5 0.019
6 0.012
Cool, thanks.
How would one do that with LADSPA plugins? 6 sine wave oscillators into a mixer, one for
each harmonic?
4092
days inactive
4094
days old
linux-audio-user@lists.linuxaudio.org
8 comments
7 participants
participants (7)
-
David Santamauro -
Fons Adriaensen -
Harry van Haaren -
Ken Restivo -
Lieven Moors -
Lorenzo Sutton -
Will Godfrey