http://www.chriswatson.net/swf/TheColourOfSound.flv

The tricky bit is the fitness function for the GA, which Bozkurt hints at using a simple of Mean Squared Error (MSE):

The fitness function is tricky, it uses an UGen I’ve coded in C for the purpose. It compares the MSE distance between complex spectrums of the source and synthesized sound and the MSE distance becomes the similarity rating therefore the output of the fitness function (lower is better). I’ll let you know when the paper describing the internals is published.
Here is a paper by Ricardo A. Garcia where he grows sound synthesis graphs (instead of parameters to fix synthesizer graphs), I use the same similarity rating formula, but discard the phase:

http://galileo.cincom.unical.it/Music/workshop/articoli/garcia.pdf

This reminded me of a question about Mel Frequency Cepstral Coefficients the other day on the SC mailing list:

To look into exactly what MFCCs are, you might want to start with
regular cepstrum. Leaving out all the details, the idea behind
cepstrum is that it’s the spectrum of a spectrum – an FT of an FT -
and it (sort of) deconvolves two convolved sounds. A sung vowel is a
convolution of source and filter – glottal impulses filtered by the
singer’s head. If you look at a cepstrum of a sung vowel, you get a
condensed representation of the general spectral envelope at the
beginning of the cepstrum, and a peak that corresponds to pitch a
little higher up (depending on frequency). In the best circumstances,
you can look at a second cepstrum of the same singer singing a
different note, and the early part of the cepstrum will look the same,
but the position of the pitch peak will be different. That’s because
the singer’s timbre is the same – only the pitch has changed. So
that’s why early cepstral coefficients are used as a feature to
describe timbre. The analyses I’ve done on voice have been the
cleanest and easiest for me to interpret. With other instruments the
“deconvolution” of spectral envelope and pitch isn’t as nice.

MFCCs are a lot different than regular cepstral coefficients. The
results of an initial FFT are multiplied against a filterbank composed
of filters that are evenly spaced on the Mel scale. The Mel scale is
an old (and now controversial) scale for pitch based on perceptual
experiments, but it does relate strongly to the Bark scale which has
been thoroughly established in relation to our hearing mechanism. If
you look at the plot of Mels against Hz on wikipedia, you can see that
Mels, like Barks, favor the low end of the spectrum. By running a
regular spectrum through a Mel-weighted filterbank, the spectral
detail of the low end is preserved much more than in the high end.
That’s because the width of the low filters is relatively small, and
the higher filters are very wide. So all that high end detail gets
averaged out. The number of filters in the bank depends on the
spacing between the filters. My Pd objects let you specify any
spacing. With 150-Mel spacing, you get 25 filters. With 100-Mel
spacing, you get 38. So regardless of your original FFT window size,
you get back far fewer points with typical Mel spacings.

To finish off the MFCC calculation: the Mel-weighted reduced
resolution version of the spectrum is subjected to a discrete cosine
transform instead of a second FT (as in cepstrum). The first MFCC
ends up being a crude indication of overall loudness, and the
remaining coefficients can be used to describe timbre pretty
accurately. Unless you’ve got specific CPU limitations, in my
experience it has been best to use all the MFCCs. They sort of
naturally weight themselves in a way anyway – as the earlier
coefficients are usually much larger than the later ones. As for
accuracy, I found a very significant improvement classifying
percussion sounds when I systematically compared cepstrum to MFCCs. I
also use BFCCs – cepstral coefficients weighted according to the Bark
scale.

MFCCs might be a good in a fitness function perhaps taking the MSE of the MFC Coefficients?

I thought this might be a good idea for audio work or samples, having children describing serious stuff.

I thought this might be a good idea for audio work or samples, having children describing serious stuff.

I have been thinking of putting together a Pd patch to augment my voice (which is terrible from years of non-use silently sitting in front of a computer). This always tends to be a solo act whenever I have seen it. I would love to see a electronic group that had a singer like this working with more experimental backing.

I was thinking that I might then use the ‘piano waves’ setup I made in Reason when I was in Czech, which really has an emotive flow to it and makes it easy to accentuate moments. Piano is easy and sounds pretty, particular against nature sounds.

Mostly of my recordings are from Urban landscapes however or of jet planes. Thomas Brinkmann and Scanner did a track called Adria in which they had a recording from inside a fighter jet on an attack run. The track, minimal tech, builds up ominously (sadly no strike on the sample so no climate).

Finally there is the obvious approach which would be to slice up the samples and integrate shorted sections. The Boards of Canada did some nice stuff in this direction on there Geogaddi album, in which they sample old documentary and play some of the samples backward

Erika took me to the Dia:Beacon yesterday and I discovered the work of Sol LeWitt! He has a series of line drawings on the walls there exploring various patters produced by moving thought all the permutations of certain lines or shapes. I feel these are similar, in a way, to how I approach experimental electronic music.

I’d like to make a series of audio works that explore audio space in the same way.

200 GUitars at the Lincoln Center

Last night Erilka took me too the Lincoln center where one of her work makes was involved in the 200 Electric Guitars performance composed and conducted (signalled?) by Rhys Chatham. It was an interesting performance in that the audience was surrounded by a U-shape of electric guitars (see pictures for one side) which where sectoned into or and had sub-conductors (signalers) who relayed the messages of Mr.Chatham to their section. Thus Chatham was able to have pan effects moving the sound from one corner to the other.

The actual content of the piece I didn’t find that inspiring and the overall effect of the 200 guitars come across as a string section of any classical orchestra. The final crescendo was more of what I expected with all sections playing (out of time with one another) and a sort of textured drone surrounding the listener. I was hoping for more of this, with the drone changing consistency in each section, perhaps rotating around, swirling…

So I thought this would be possible with SuperCollider and using many voices (perhaps 200 even!). I recently read a tutorial on how to recreate the THX audio logo which is made up of 30 voices out of sync that converge. I’d like to combining the two ideas to create a 3D audio composition in which various drone elements would move around the listener and converge in-and-out. This wold only be possibel by computer I think as a lot of the limitations in Chathams composition where due to the ability to control the guitars.

Overall a delightful experience!

A bit like that Ricardo Villalobos minimal experimental thing with the harp but flipped with the backing being abstract (instead of drums) and the melody [layer] being more concrete.

Perhaps a traditional blues riff or even a cover with some lyrics. I was thinking that I might be able to add high pitched vocals if I put my voice though some sort of filter to make a wale out of it (until I get my voice back).

I found a very cool drummers blog by some LA session drummer who pumps out a drum loop a day and talks about drumming technique in general. Has really helped scope out the head space of drumming theory for me, and I constantly keep checking back so I’m always thinkging about it.

Things I have learn for composition:

• Changing the base drum sound without it seeming like a different track all of a sudden.
• Really can’t have more than one bass drum sound (perhaps if one was raddically different, say a techno sound and one was a normal kit sound ?).
• Its nice to mix constant synth drum backbeat with a more organic sampled drum sound over the top of it, perhaps just playing the high drum sounds.
• The action or character in a dance track is all in the top part of the drums, bass alone is boring and the high parts should work with the melody.

Thats all I cn think of right now.

SOLID CREATIVE PROCESS

While working on this new track I cam to the idea to work on one element at a time and when I had it to my liking, combined it with a second and worked the two together, getting ideas for recording the two together, then laid this down, and then finally introduced the third element. Like mixing a cake, bringing in new elements a little at a time once the existing ones are combined well. Its a solid process but slow and can be draining.

TODO

I think I need to practice more improvisation and live recording as I feel a bit scared of doing anything live and I feeling it would speed up things and keep it fun (and require less energy).

Also, I’ve rendered out all the stage 3 tracks I’ve made and imported to iTunes so I can listen to them mastered down and see how I feel about them. Already after one hearing I feel I know things that need to be changed in some tracks and see what it perfect the way it is (well, some mastering couldn’t hurt on them all).

The question really is what gives some rthyum some swing, or funk and others nothing? I know that the basic swing beat, alternating between two sounds every 4 beats gives the maximum swing. The two seperate sounds each create their own space and the mind moves from one to the other with enough time between them, back and forth like a swing. The more sounds you bring in the more the mind jumps. Things to experiment with here might be:

• Many low sounds verses many high sounds? I suppose the depth and strength would make some stand out in a jungle of sounds. Lower sounds being more rhythmic and higher being more melodic (?)
• The maximum and minimum (i guess 2?) number of sounds. How many can the mind track in rhythms before it becomes just noise? I guess the types of sounds, how close together they are makes a difference here.

Today I hear some Spanish sounding music and it was using a very tight change in velocity while appegiating like mad over something. It was an organ sound and had a pop’y feeling to it.

I was using a mix of classic rock style drumming on the stead state and some weird electronic stuff which I messed with even more by playing with the LFO on Dr:Rex, applying it to the filter and the OSC.

I then tried to lay down a choppy, yet fat, bass over the top. I need some more experience making the bass work with the chops in the drums. Perhaps the Groove mixer ould help here a bit(?).