Palimpsest (or: How To Destroy Sounds)

[tetraplan]

In the thread following the posting of my latest track, the following bit of dialogue took place:

//looks up "palimpsest"
Hmmm. Never thought of sounds as palimpsests. Are you referring to the atomic recycling of sounds in the granular process?

Sort of, but more the layers of remnants. When you see old hand written manuscripts with stacked erasures, there's this sense of activity in all the bits and pieces.

Well, Steve gave me a great idea.
How does one scrape layers of sound off an audio-file?
The idea is that something new comes out, recycling as it were.

If I found a way, I promised Steve, I would let him know.
I figured that more people would be interested. Maybe not, but one never knows.

The solution was right in front of me. The spectral noise reduction found in tools like Audacity, Goldwave and Acoustica (well, most editors, really) do just that: remove layers of sound. And in a messy way, too, when over-applied.

I have just finished stage one.
My idea was this: I will use one soundfile as noise-source, and another to be horribly mangled. A bit like convolution, as suggested by Steve, only really destructive. (mwa-ha)

I downloaded the promo.mov from the gunbuilders site that was posted by Glassback a couple of days ago (thanks), exported the wave-file and loaded this into Audacity.
On the next track, I loaded the raw data from the *.mov. I used this as noise-profile.
I applied the profile to the soundtrack, and the end-result was the sound of 100 tin birds chattering. Really cool!
I did pitch it down by 2 octaves because the sounds were so high-pitched they were barely audible.

Example 01


Experiment two was the reverse: used the sound-track as profile and applied this to the *.mov.
Weird whistly sounds wwere the result.

Example 02


For experiment three I reversed one instance of the soundtrack and used this as profile. I applied this to a normal version of the soundtrack.
Nice wobbly spectral shifting. Like a shortwave-radio signal coming from under water.

Example 03


Phase two will be adding a new sound to the palimpsest. Not sure if I sould somehow FFT this on top of the new sound or if I should just use the new sound as source in sampling or resynthesis.

Before I start that, I want to try an alternative way of erasing sound. I dug out Hog and found that it does frequency-extracting. This, too, could be an approach. Will try this later.

//edit:
It just occurred to me that it would have been a better idea to post this in the Music Cafe, with sounds.
Oh well, late, tired, &c.
//edit 2:
Links to examples added.

Groet, Erik

[Sicklecell666]

post it in 'everything else' or the 'samples' thread..

I wanna hear this when it's ready!

[tetraplan]

post it in 'everything else' or the 'samples' thread..

I wanna hear this when it's ready!

OK.

Groet, Erik

[tetraplan]

OK, links added to original post.
These are just the results of phase one. Will post later results when ready.

Groet, Erik

[hao nao]

cool experiments, Erik

[mystahr]

Creating through detracting instead of adding; quite interesting Erik

[skybax]

nice!

[mystahr]

Just heard only now since I was listening to something else when I stumbled on this thread.

This sounds superbly interesting

[The Chase]

in adobe audition, if you apply the noise reduction to a wave, using the same wave as an example for the noise reductor, you will hear that very trippy sound that Aphex Twin uses a lot in his music, except that he has something that allows him to tweak it in real-time.

Here's my attempt (2 examples in one small mp3)

[Sicklecell666]

that's pretty cool tetra-plan..

I dig your sample source

They should use your effected audio for the next DOD presentation; it lends a bit more wieght to DREAD..

[shamann]

Interesting stuff, Erik. Fun concept to filter the audio with the video.

I'd done similar with the FFT filtering in Cool Edit, but hadn't really considered it along the palimpsest lines. One thing about palimpsest is that they're often remnants from previous draughts and errors.

Still not sure if this is the perfect solution, but it is better than convolution. Tried out some things and it tended to make it very resonant more than anything else. Only thing with FFT filtering, need to figure out a way of fading it in to the background effectively (volume reduction I suppose, but the background sound need to be an imperfect reflection of the forefront audio).

Here's a test I did with just with mixing of various versions of the same phrase and some bitcrushing/distortion. Perhaps a bit too literal, not really thinking fully on digital terms.

You've given me some ideas to test out now.

Cheers,
Steve

[tetraplan]

Thanks for the comments, everyone.
It's fun to do resarch like this.

In adobe audition, if you apply the noise reduction to a wave, using the same wave as an example for the noise reductor, you will hear that very trippy sound that Aphex Twin uses a lot in his music, except that he has something that allows him to tweak it in real-time.

Here's my attempt (2 examples in one small mp3)

I think Aphex Twin uses Max/MSP for this stuff. Has a lot of FFT-based externals.
I think this could be done in PD, then, as it is based on Max.

Hmmm. I wonder what happens if you use the same *.wav, only pitched up or down.

I dig your sample source

They should use your effected audio for the next DOD presentation; it lends a bit more wieght to DREAD..

Yes, I am looking for specific sourcefiles like this (like NOAA-data that was used for 975 Millibar), because mangling for the sake of mangling doesn't do much for me. I like working in a way that's a bit more conceptual/theme based (1).

You should see what happened to the video once I exported it as uncompressed *.avi. Winamp didn't know what hit it.

Interesting stuff, Erik. Fun concept to filter the audio with the video.

I'd done similar with the FFT filtering in Cool Edit, but hadn't really considered it along the palimpsest lines. One thing about palimpsest is that they're often remnants from previous draughts and errors.

Still not sure if this is the perfect solution, but it is better than convolution. Tried out some things and it tended to make it very resonant more than anything else. Only thing with FFT filtering, need to figure out a way of fading it in to the background effectively (volume reduction I suppose, but the background sound need to be an imperfect reflection of the forefront audio).

Hmm, yes. I see the results of my experiments as something that would have to form the base of a new sound, almost like the foundations of Inca-temples became the base for churches in the Andes (at least, that's how i visualise them right now).

Here's a test I did with just with mixing of various versions of the same phrase and some bitcrushing/distortion. Perhaps a bit too literal, not really thinking fully on digital terms.

A bit like a fractured shadow of the source. Interesting.
I was wondering how I could meld the different sounds into one, but maybe layering/mixing would do the trick.
Maybe I've been thinking too digitally.

You've given me some ideas to test out now.

Cool.
Looking forward to the results.

Groet, Erik

(1) not that the source remains in any way recognisable.
It's a bit like one of my teachers said: "When you decide to make a work of art that consists of a pile of books, every element of it should be an actual book.
If only the top element is a book, because that would be the only one you could open anyway, and the rest just looks like books, it's just decoration."
A bit like the work of Thom Puckey in our local museum: there are 3 iron globes buried in the ground. You'll never be able to see them, but they are there, arranged according to plan. Even if you can't see them, it's important that they are actually there as part of the piece.

[donkey tugger]

I remember Guy Sisgsworth going on about using a similar technique a few years ago in Sound in Sound;

http://www.soundonsound.com/sos/mar01/a ... 6986912533

Keep at it, you could end up producing Madonna too!

[tetraplan]

I remember Guy Sisgsworth going on about using a similar technique a few years ago in Sound in Sound;

http://www.soundonsound.com/sos/mar01/a ... 6986912533

I'm fascinated by the qualities of time-stretching. Sometimes I've time-stretched parts and then time-stretched them back again, so it's not to change the length, it's just to get that weird shifting of the harmonics that time-stretch puts into a sound. I discovered a wicked thing the other day using the Digidesign Noise Reduction plug-in. When you drive it too hard, it starts, again, to mess with the harmonics of the sound you're putting into it. You're meant to show it an example of the noise it's trying to remove, and then you show it the sound it's got to remove it from. If you deliberately show it a completely unrelated sound, so it's looking for those harmonics in this other sound, you can create really weird harmonic resonances.

Heh. Drat.
Almost to the letter. This is the best evidence of convergent evolution.
I wonder what he does with the results- digital oddness on a Madonna track?

He has a DMX, too. The bastard.

Keep at it, you could end up producing Madonna too!

Uh-oh.
If you ever see me sporting a piece of red string around my wrist, slap me.

Groet, Erik

[cptgone]

looks like you've been having fun