Dawesome Kontrast. Wavetable synth. Coming Autumn 2025.

[foosnark]

One thing that I think might be confusing people, is that the X and Y directions of the image being scanned don't actually mean anything. With a spectrograph, it's frequency vs. time. With a typical wavetable, it's level on the Y axis and phase on the X axis ("time" comes from the oscillator frequency).

But with this synth it's just a 2D map, where X and Y don't have any particular meaning. (*) The brightness of each pixel represents a level. A scanner travels over it, reads the pixel brightness and translates that to level. The timbre is going to be dependent on the shape/complexity of the path, and the smoothness and density of changes in brightness where the scanner travels over it.

This might work two possible ways that I can see:

- direct scanning. Frequency is dependent on how fast the scanner moves. It just flies over the map, bright spots producing peaks in the wave and dark spots producing troughs.
- the scanner fills in a buffer, which is played like an ordinary wavetable but its data keeps getting updated. Slower scanner motion means slower change in timbre but not lower frequency. (The problem I see here is a huge potential for discontinuities, where you're scanning a very smooth map but the start/end points don't match up. So maybe there's something that automatically chooses a buffer length with a closed loop. Or, maybe this isn't a "problem" at all but part of the characteristic sound...?)

Anyway, some examples:

1. The map is a checkerboard. As long as your scan travels far enough, this is going to create pulses of some kind -- maybe a squarewave, maybe PWM, maybe weird pulse trains depending on the shape and complexity of the scanner's path. (But a really tight circle that stays 100% over a dark or light area would be silent.)

2. The map is a simple gradient, bright at the top and dark at the bottom. Effectively, only the Y position of the scanner matters here. If the scanner loops in a circle, it produces a sine wave. (Again, assuming the start and end points line up or it's scanning directly). If it zigzags linearly up and down, it's a triangle. If it starts at the top, goes down and then jumps back up, it's a sawtooth. With more complex smooth curves (the flower petals or spiraling) I'm guessing it's a bit like additive synthesis with a few harmonics. If the path doubles back on itself with acute angles, it's going to be more like wavefolding.

3. The map has more bumps, ridges, edges, noise... these are going to result in more and stronger upper harmonics, edgier waveshapes with a brighter sound.

(*) I guess if you import a sample, it will render it with time on one axis. But it isn't still time afterward, unless your path just linearly travels in the same direction.

[whassup]

I am stunned.
This makes so much sense and seems to be executed so well.
And good looking obviously.
I so would love to get my hands on this...

[kraster]

One thing that I think might be confusing people, is that the X and Y directions of the image being scanned don't actually mean anything. With a spectrograph, it's frequency vs. time. With a typical wavetable, it's level on the Y axis and phase on the X axis ("time" comes from the oscillator frequency).

But with this synth it's just a 2D map, where X and Y don't have any particular meaning. (*) The brightness of each pixel represents a level. A scanner travels over it, reads the pixel brightness and translates that to level. The timbre is going to be dependent on the shape/complexity of the path, and the smoothness and density of changes in brightness where the scanner travels over it.

This might work two possible ways that I can see:

- direct scanning. Frequency is dependent on how fast the scanner moves. It just flies over the map, bright spots producing peaks in the wave and dark spots producing troughs.
- the scanner fills in a buffer, which is played like an ordinary wavetable but its data keeps getting updated. Slower scanner motion means slower change in timbre but not lower frequency. (The problem I see here is a huge potential for discontinuities, where you're scanning a very smooth map but the start/end points don't match up. So maybe there's something that automatically chooses a buffer length with a closed loop. Or, maybe this isn't a "problem" at all but part of the characteristic sound...?)

Anyway, some examples:

1. The map is a checkerboard. As long as your scan travels far enough, this is going to create pulses of some kind -- maybe a squarewave, maybe PWM, maybe weird pulse trains depending on the shape and complexity of the scanner's path. (But a really tight circle that stays 100% over a dark or light area would be silent.)

2. The map is a simple gradient, bright at the top and dark at the bottom. Effectively, only the Y position of the scanner matters here. If the scanner loops in a circle, it produces a sine wave. (Again, assuming the start and end points line up or it's scanning directly). If it zigzags linearly up and down, it's a triangle. If it starts at the top, goes down and then jumps back up, it's a sawtooth. With more complex smooth curves (the flower petals or spiraling) I'm guessing it's a bit like additive synthesis with a few harmonics. If the path doubles back on itself with acute angles, it's going to be more like wavefolding.

3. The map has more bumps, ridges, edges, noise... these are going to result in more and stronger upper harmonics, edgier waveshapes with a brighter sound.

(*) I guess if you import a sample, it will render it with time on one axis. But it isn't still time afterward, unless your path just linearly travels in the same direction.

I'm not sure that the visualisation here is anything more than just a visualisation of an underlying traditional frame based wavetable. (e.g. 256 frames with 2048 samples per frame)

As Peter said above the visualisation is not a spectrum. Instead of the traditional raster based representation like you get in serum or Phaseplant you have something that looks more like a relief map.

I could be wrong but my impression is that instead of the traditional line based scan of wavetables you have different shapes that can dynamically straddle different frames of the wavetable so the instantaneous amplitude values of the positions of the shape at a given point in time make up the output waveform at that point.

This makes it more like a much more complex and dynamic version of phase distortion

[Old Norse]

But then on the product page, he advises us to go out and take pictures of "bathtubs, structures and radiators" to use as wavetables. So there's that.

[foosnark]

I'm not sure that the visualisation here is anything more than just a visualisation of an underlying traditional frame based wavetable. (e.g. 256 frames with 2048 samples per frame)

As Peter said above the visualisation is not a spectrum.

Just in case you misinterpreted me, I did not say it was a spectrum, in fact I said it was not

But it's not really a traditional wavetable either. In those, the X axis is ALWAYS traversed from left to right at audio rate, representing the phase within the cycle, and the Y axis can be optionally set by some modulation source.

With Kontrast there's a scan position that is traveling on both the X and Y axes, according to its own algorithmic behavior. The X and Y axes on the map itself don't have any particular meaning, it's just a spatial arrangement. Things happening over time are dependent on the scanner movement, not their location on the map.

Anyway, this is all semantics and I'm pretty sure it will be quite intuitive once we can actually play with it.

[sandandpaint]

I'm not sure that the visualisation here is anything more than just a visualisation of an underlying traditional frame based wavetable. (e.g. 256 frames with 2048 samples per frame)

As Peter said above the visualisation is not a spectrum.

Just in case you misinterpreted me, I did not say it was a spectrum, in fact I said it was not

But it's not really a traditional wavetable either. In those, the X axis is ALWAYS traversed from left to right at audio rate, representing the phase within the cycle, and the Y axis can be optionally set by some modulation source.

With Kontrast there's a scan position that is traveling on both the X and Y axes, according to its own algorithmic behavior. The X and Y axes on the map itself don't have any particular meaning, it's just a spatial arrangement. Things happening over time are dependent on the scanner movement, not their location on the map.

Anyway, this is all semantics and I'm pretty sure it will be quite intuitive once we can actually play with it.

Initially I thought the scan was a trajectory along a path but the dev said the image is scanned left to right with a vertical playhead. In which case there can be multiple Y values for any X and that has to be resolved somehow

[kraster]

But then on the product page, he advises us to go out and take pictures of "bathtubs, structures and radiators" to use as wavetables. So there's that.

Image to wavetable isn't new though. Serum has had it for years. As have metasynth and others.

Horizontal axis is frequency from low to high . Luminance is the amplitude of each frequency.

And the vertical axis represents the frames.

It uses an inverse FFT to convert the image row values to a waveform.

You still end up with a wavetable.

[sandandpaint]

But then on the product page, he advises us to go out and take pictures of "bathtubs, structures and radiators" to use as wavetables. So there's that.

Image to wavetable isn't new though. Serum has had it for years. As have metasynth and others.

Horizontal axis is frequency from low to high . Luminance is the amplitude of each frequency.

And the vertical axis represents the frames.

It uses an inverse FFT to convert the image row values to a waveform.

You still end up with a wavetable.

this makes more sense to me than a direct read where there can be more than one Y value for every X

[PetVor]

Initially I thought the scan was a trajectory along a path but the dev said the image is scanned left to right with a vertical playhead. In which case there can be multiple Y values for any X and that has to be resolved somehow

The scan is a trajectory along a path!

This includes traditional wavetable synthesis as a special case: just make the scanline a straight vertical line. Moving this vertical line left and right would then correspond to moving the slice position in a traditional wavetable synth.

In traditional wavetable synthesis you then have one or two modifiers that will modify the wavetable. In KONTRAST, instead of modifying the wavetable, you modify the form of the slice: it can have almost arbitrary shapes.

[PetVor]

[...]
Horizontal axis is frequency from low to high . Luminance is the amplitude of each frequency.
And the vertical axis represents the frames.
It uses an inverse FFT to convert the image row values to a waveform.
You still end up with a wavetable.

When you import audio or a wavetable in Kontrast:
* the horizontal axis of the wavetable represents the time (or slice position)
* each column is one slice as a waveform
* white values are mountain tops of the waveform
* black is valleys in the waveform

Here is a page from the manual:

[PetVor]

But the whole point is: for sound generation you can forget about time and just treat the audio as a 2D map. You are no longer limited to slices of audio and one axis to move, instead you can move around freely!

[Shabdahbriah]

^^^ Kinda like moving a cross-hair (or fixed lens POV) around a 'bump-map' ?

[kraster]

But the whole point is: for sound generation you can forget about time and just treat the audio as a 2D map. You are no longer limited to slices of audio and one axis to move, instead you can move around freely!

I was sold when I read "New" and "Dawesome" in the same sentence.

Is there an ETA?

[PetVor]

Is there an ETA?

Release is now planned for 3rd October

[morphex]

But the whole point is: for sound generation you can forget about time and just treat the audio as a 2D map. You are no longer limited to slices of audio and one axis to move, instead you can move around freely!

I think this is a great idea. I especially like your emphasis on designing instruments that we can spend more time performing with, and less time on tedious programming and menu-diving (though, that has its place).

The overall Dawesome aesthetic is obviously an important factor, and one that has been very well thought out. I personally like your move away from standard UI's to ones that feel more akin to using a piece of artwork.

Also, having access to expressive, performance-oriented features such as MPE functionality, is bringing the artist and the digital tool closer together and more in sync. That tighter integration could foster more intuitive playing.

And since there's a good chance that will result in a more fulfilling experience for the performer, that vibe will come across as better, more emotive music. These are all intriguing possibilities that my optimistic side sees in Kontrast, as well as the entire Dawesome lineup.