The Eat at Joe's Kawai K5000 Message Board Digest
Generating Noise Using ADD
------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------
Close to the noise
Thursday, 28-Jan-99 08:23:12
194.172.230.108 writes:
Noise with additive synthesis? - Impossible! That's why Kawai gave us those PCM
snippets.
But if we tried... couldn't we expect to get at least some nice weird sci-fi sounds?
So - what do you think? How close to noise can we get? With which tricks?
( g r o h @ i r t . d e )
Jens Groh
------------------------------------------------------------------------------------------
Re: Close to the noise
Thursday, 28-Jan-99 10:07:01
192.28.2.49 writes:
Check out the OHH patch in the archive and the discussion about it.
Doing it without random LFO functions would be a challenge!
leiter
------------------------------------------------------------------------------------------
Re: Re: Close to the noise
Thursday, 28-Jan-99 21:50:48
209.215.4.80 writes:
Hmmm, neat challenge. I made strange bass drum and "very loose" tom tom sounds in
the past using a sine LFO. But for real noise, I'd just HAVE to delve into
properly insidious manipulation of the upper 64 harmonic series...
Terry
------------------------------------------------------------------------------------------
Re: Re: Re: Close to the noise
Thursday, 28-Jan-99 23:47:33
207.104.158.184 writes:
ya thats what i was thinking. every time i mess with the upper harmonics i usually
end up with some painfull noise... does anyone have any really usefull upper
harmonic patches? maybe ill check the archive.
andrew
------------------------------------------------------------------------------------------
Re: Re: Re: Re: Close to the noise
Friday, 29-Jan-99 05:43:13
194.172.230.108 writes:
Check out my 'BioReact' patch.
Jens Groh
------------------------------------------------------------------------------------------
Re: Re: Re: Re: Re: Close to the noise
Saturday, 30-Jan-99 12:46:38
199.199.157.58 writes:
Cool patch--it took awhile to figure out where the sound was coming from. The
patch supplies it's own rhythm, too. You have a lot of "patch=track" patches!
leiter
------------------------------------------------------------------------------------------
Re: Re: Re: Re: Re: Re: Close to the noise
Sunday, 07-Feb-99 07:13:40
194.172.230.108 writes:
1. - leiter wrote:
[ Cool patch--it took awhile to figure out where the sound was coming from.
The patch supplies it's own rhythm, too. You have a lot of "patch=track"
patches! ]
Thanks - but the king of "patch=track" patches is Matt Jasper! It's a mystery how
he makes the sounds evolve over minutes!
2. - I'm still trying to get more noise out of additive than your hi-hat patch -
without much success. My idea was to use sparse harmonics selected to be as
"inharmonic" as possible, and the ring modulator to drive it away from tonality.
3. - Sorry, I was wrong saying that "BioReact" uses high harmonics (though it's
no harm to switch them to "HI").
( g r o h @ i r t . d e )
Jens Groh
------------------------------------------------------------------------------------------
Re: Close to the noise
Saturday, 30-Jan-99 14:27:22
199.199.158.38 writes:
Frequency spectra for real instruments seem to have fairly broad peaks. If it
isn't an artifact, it means that there's a "cloud" of slightly out-of-tune tones
around each harmonic. In pipes, this might be because the frequency response of
the sound generator has a gentle slope between maxima and minima. (The maxima and
minima fall on odd or even members of the harmonic series.) I'm thinking
this may account for the "breath noise" of pipes.
I'm sending a patch to Kenji and the list called "Cloud Pipe". H16 and H48 are
taken as the first two harmonics of a triangle wave. (So, even with the max
adjustment, it plays two octaves high.) The harmonics above each of these are
added to make a "cloud". Also, a second ADD, sharp by about 24 cents, puts in
more "cloud" harmonics.
There are no PCM's and no random LFO's but I think I hear "breath noise" (with
or without the Celeste effect). Let me know what you think.
leiter
------------------------------------------------------------------------------------------
Re: Re: Close to the noise
Tuesday, 09-Feb-99 09:20:38
194.172.230.108 writes:
One of your most ingenious patches!
Here is how I would explain how it works:
Basic thought: Generate roughness through modulation. (I can think of both AM and
FM as candidates.)
Theory says: Modulation causes side-bands to occur.
Revert the argument: Side-bands cause modulation!
Side-bands seem to be interpreted by the ear as (narrow-band) noise. There
is a noise impression even if the signal is actually periodic, as long as
the periodicity lies in the subsonic range. (1/16 of the 'fundamental'
frequency in this case.)
The modulation period is mainly determined by the frequency spacing of the
harmonics. (The 'real' fundamental does not exist.) The roughness sensation
will probably be pitch-dependent: Too slow periodic modulation will be perceived
as rhythm, a too fast one will make you hear a buzz.
Further ideas:
CloudPip uses only the upper side-band. Why not use both side-bands?
Will vibrato (FM) smear the spectrum even more?
Do the optimum side spectra possibly follow the so-called masking curves
(known from psychoacoustics)?
This might prevent perceiving side-bands as separate tones, yet increase
the noise energy...
Jens Groh
------------------------------------------------------------------------------------------
Re: Re: Re: Close to the noise
Tuesday, 09-Feb-99 21:14:48
199.199.158.16 writes:
>CloudPip uses only the upper side-band. Why not use both side-bands?
Not sure why I didn't--just added it and it does sound better; "noisier".
I can send the new version if anybody wants.
>Will vibrato (FM) smear the spectrum even more?
It's in there already, a little fast vibrato on each ADD.
>Do the optimum side spectra possibly follow the so-called masking curves
(known from psychoacoustics)?
>This might prevent perceiving side-bands as separate tones, yet increase
the noise energy...
That could be. I went looking for a chart, but I haven't found anything that
would give me numbers yet.
Did find some cool stuff on temporal masking, tho. According to a science
encyclopedia, masking persists after the masking note is gone. Also, masking
can work "backwards in time", i.e., a later note can mask a previous note. I'm
skeptical, but it could be. The article didn't say how much previous, but I
assume it's very short. Seems like this would mean that, for attack times below
a certain limit, a note will mask it's own early-attack and be perceived as
instantaneous.
leiter
------------------------------------------------------------------------------------------
Re: Re: Re: Re: Close to the noise
Wednesday, 10-Feb-99 04:51:24
194.172.230.108 writes:
Leiter, you wrote: "Also, masking can work "backwards in time", i.e., a
later note can mask a previous note. I'm skeptical, but it could be. The article
didn't say how much previous, but I assume it's very short."
Yes! The phenomenon comes from the nervous processing which is faster for loud
sounds than for soft ones. All sensory events (light, sound, pain...) which are
strong may indicate an alarm situation and are thus processed faster. And they
mask weaker events. So the attack is perceived a little earlier than the rest.
The order of magnitude for the pre-masking time is about 1 msec, for post-masking
about 10 msec, as far as I remember. (I worked on that theme a few years ago.)
Jens Groh
------------------------------------------------------------------------------------------
Re: Re: Re: Re: Re: Close to the noise
Friday, 12-Feb-99 15:56:45
192.28.2.16 writes:
>The phenomenon comes from the nervous processing which is faster for
loud sounds than for soft ones. All sensory events (light, sound, pain...)
which are strong may indicate an alarm situation and are thus processed
faster. And they mask weaker events. So the attack is perceived a little
earlier than the rest.
That reminds me of a time when I was standing near a door when it was slammed
shut by a gust of wind. I jumped, as a reflex, but it seemed to me that I had
jumped before I heard the sound. The "ear>reflex>awareness of jumping" path must
be shorter than the "ear>awareness of sound" path.
>The order of magnitude for the pre-masking time is about 1 msec, for
post-masking about 10 msec, as far as I remember. (I worked on that theme a
few years ago.)
Thanks for the references. It seems like complete masking is more interesting
from a data compression standpoint than a musical standpoint. But "fusion", as
one article called it, is interesting. A drum "flan" may be an example. I've
tried some experiments with the "Snare Attack" PCM and the delay parameter of
the K5k. For a prior hit, I haven't heard any masking or fusion yet; one notch
of the delay parameter may be too long.
For subsequent hits, though, I think I've heard some fusion, i.e., a change in
character of the sound without hearing two seperate hits, but I'll have to play
around with it some more.
Another thought, compressing a track with strong kick and snare is similar to
masking.
leiter
------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------