Make Noise - New Spectraphon Module

[vurt]

starts with a single sample.
then a sam version of the same sample joins in.
then an array of the same sample on the other channel. (then some echophon is added to the bass drone from this)

then delay and such...

Sounding good! Watching live at the moment

cheers

not too bad for a second of a crossfading cymbal and sitar

[foosnark]

OK, I learned something while experimenting just now, and it's best to show the results in audio form.

1. This starts off with Rings (in modal mode), dry, triggered by a steady clock.

2. Then Rings through SAM. Partials was max, FM off. Focus was very low. Slide and Freq were adjusted by ear to sound like a sort-of match, not a perfect one -- the goal wasn't perfect replication in SAM but trying to see how SAO and SAM differ.

I recorded an array using the default clock rate -- no attempt to sync anything to the tempo that triggered Rings.

3. Next we hear SAO, with a linear ramp into Slide. The envelope is all wrong, but more importantly the timbre sounds very "flattened." As it turns out, this is not caused by normalization during recording, but by the interpolation that Slide does.

4: Now it's SAO clocked by Shapeshifter. The timbre match is much closer to SAM, maybe a little changed by the normalization. But the sync issue becomes obvious. You can't set the start and end points, you just have to iterate through the whole array.

5: And last, SAO with a linear ramp into Focus. The envelope is a little weird and we hit a point where it loses the movement that it should have.

https://starthief.net/stuff/spoon_experiments.mp3

So really, the problem for replicating results isn't that something bad happens to the data when it's captured. It's that both clocking the array during capture, and moving through it during playback, can be critical for some kinds of material. But also awkward.

Using a clock to play back would have worked better if I'd managed to sync the length of my recorded note with the array length, which means being careful with clocking. Or maybe editing the spectra file on a computer with some kind of tool to trim the excess...

[vurt]

ah, so silence doesn't effect the array?
i know it has no spectrum, but is it truly silent if something is plugged in? was my thinking.

that's why ive been looping everything with no gaps

[vurt]

ive just discovered...
in sao mode, even if there is nothing plugged in to the input from sam mode/array creation, the gain still has an effect on the array playback.

[foosnark]

AHA!

I think I have figured out that Slide is subdividing the array into 8 steps, by skipping over some entries, and then interpolating between those steps that it picked.

E.g. if you have 32 steps in your array, moving Slide will go between steps 1, 5, 9, 13, 17, 21, 25, 29. Hit Clock once, and instead Slide goes between steps 2, 6, 10, 14...

If you want to morph between two sounds, manually clock the first one four times, switch, then manually clock the second one four times. That will work fine.

If you create an array with 16 steps but every other step is silent, you can move through Slide without ever hearing silence. But turn Focus (or hit Clock) and you reach silence, and then you can move Slide all over and it's still silent.

And if you create an array with less than 8 steps, you can't move through it at all.

Focus interpolates between individual adjacent steps, while Slide gets you some blends that you couldn't necessarily get in other ways. But the combination of things means there's no way to really navigate the entire array with CV. I'd like a mode where Focus has the full array range instead of just part of the range.

I'm going to confirm the details with Tom Erbe (and ask a few other things) but it's starting to make a LOT more sense.

[vurt]

i had an idea it would be something like that, but didn't reckon on a minimum. just assumed 2 points would be the beginning and end of slide, with focus being a fine tune kind of.
but what you are saying makes more sense, especially as we are dealing with a fixed maximum.

thanks that's given me some ideas for my next experiment.

[vurt]

last night, i spent far too long, listening to tiny tim, through sam mode for it to be healthy.

[foosnark]

I finally pulled the SD card and looked at saved arrays. They appear to be 16-bit mono WAVs at 48K. Each spectra in the array is 64 "samples" long, with the amplitudes of the harmonics as levels of the sample. Except that it seems to be using unsigned integers where WAV expects signed, so big numbers show up as negatives in a WAV editor.

spectra.png

This is a what an array full of squarewaves looks like -- strong fundamental, then progressively weaker odd harmonics and no even harmonics. Doesn't sound pretty if you try to play it as a regular WAV file

I tried to synthesize my own saw pluck by editing a WAV file and saving it to the card, but it didn't really work out. Maybe something in the WAV header lists the number of spectra in the array and it didn't match, or some other picky thing. I'll ask But frankly that kind of editing is pretty tedious. Maybe somebody will build something more like WaveEdit but for spectra arrays (or even, convert from WaveEdit wavetables to arrays!)

[vurt]

i went 7, 8, 8. this is a sitar, cymbal and tiny tim going ahhhah.
then just using the clock to step through those bits.
get quite a rezzy sound from the cymbal i think it was?

(only a minute or so and no fx )

[vurt]

tried scanning through a drum loop using the clocking method, with wogglebug at maximum!
left it clcked like that, switched to array!

sounded like shite
but!
then as i didnt want to waste the experimental array straightaway...
slowing the clock down did lead to some interesting tones.
with have to try it with some other looped instruments, see how it goes

[vurt]

round two!
ambient loop, im pretty impressed here!!!



again, clocked as fast as possible, you can hear how long the loop lasts compared to the onboard clock, and i bring in the loop so you can hear how reasonably close it is!

[foosnark]

You guys should ask Tom Erbe about it, hes a super nice guy.

I kind of bombarded him with questions and a couple of possible bugs or weird anomalies. He said he'd have to get back to me about the rest later, but:

"i can tell you that it is working in the time domain, no ffts involved."

So there we go, that's how it's doing its thing without noticeable latency.

[vurt]

not sure what ive pressed, but my fm on side b doesnt seem to do anything?
im sure it did yesterday?

ive read the manual, but the fm bus bit is not really helping, unless im missing something?

[foosnark]

Huh, that's odd.

[vurt]

yup, if i can't work out whats happened tonight, ill bring it up in chat, or maybe just email support.

my worst nightmare is having to return it and get a new one, those floating rails