Zebra2 SpectroMorph & SpectroBlend Explanation

[irond13]

From the manual (& playing around) it's pretty clear that X-axis in this mode corresponds to frequency.

What's not clear, is the exact relationship to input (pitch) & output frequencies. Are the frequencies referred to on the X-axis relative or absolute? Also, I noticed that the Y-axis in SpectroBlend mode seems to have negative and postive values - how exactly does this affect things?

[Sendy]

The X axis in the spectral modes corresponds to harmonics, i.e. multiples of the note you're playing (the fundamental). That's why when you play different notes the whole thing remains consistent from note to note. Negative amounts of a harmonic represent inverted phase for that particular sine wave/harmonic.

[pdxindy]

Edit: What Sendy said...

[TomTwohy]

You're making wave shapes, so it's the harmonic spectrum you're shaping. The up and down is (I think) literally how the speaker cone moves back and forth. Amazing stuff!

You can make "fizzy", noisey sounds by using lots of complexity, smooth sound by making smooth shapes... fundamentals are towards the left, the upper harmonics are to the right.

I find the more "sine" like sounds (pure notes) come from the extreme left (try using just one or two bars), the right is so high and strange its almost psycho-acoustic. The upper range is where the human ear picks up details... the subtle difference between a tea-pot and a train whistle is determined up there.

The more complex the sound, the more complex the wave... but figuring out what is needed from such a broad spectrum is tricky indeed.

You can hear these effects if you raise one bar at a time and hit the note. You may find blending harmonics is tedious work, but also offers AMAZING results.

[irond13]

Many thanks! These explanations, together with a re-reading of the description from the manual and a brief refresher on harmonic theory has brought a lot of clarity.

So basically, the X-axis is the frequency offset, starting at 0, relative to the input pitch/frequency AKA fundamental frequency. The maximum frequency offset (i.e. the right-most value) is dependent on the fundamental frequency.

In SpectroMorph mode, the frequency offset values are logarithmically distributed across the X-axis, while in SpectroBlend mode they are linearly distributed across the X-axis. This is actually implied in the manual, where it says that *harmonics* are scaled logarithmically and linearly, respectively, across the X-axis. It turns out that this is equivalent, because the frequency difference between any harmonic N and N+1 will always be a constant value, which by definition, is the frequency of the fundamental.

The distribution of musical notes and octaves, however, is different. In SpectroBlend mode, for example, octaves are *not* distributed linearly across the X-axis (obviously deriving from the fact that they aren't linearly distributed across the frequency/harmonic spectrum either). Octaves are distributed using the formula F * 2^N, where F is the fundamental frequency. This means that the 6 octaves available in SpectroBlend mode should be distributed across the given 128 bars as follows: 2 -> 4, 4 -> 8, 8 -> 16, 16 ->32, 32 -> 64 and lastly 64 -> 128.

Now the next part of the journey is to figure out how to manipulate these harmonics accurately..

[TomTwohy]

Yup... sweeping though a series of hand made waves using MSEG or the ModMartix is VERY exiting stuff!

Setting the details of such a sweep UP... depending on how "accurate" you need to be... can be tedious

This feature is SO powerful and unique! I LAUGH at lesser synths now... the only thing better would be a line input and sampling on top of it all (retaining the manipulative aspects).

External spectrum analysis could be handy, especially if you can set up "snapshots" of the waves you are modeling... but I just use my ear and experimentation... not pretty, but neither is my music!