WaveTables and TransWaves and GeoBlend… oh My!

A wavetable is a sequence of fixed length single cycle audio waveforms of equal frequency. Multiple complex sounds can be generated from a single table by playing through the waves in a different order. Wolfgang Palm developed the powerful PPG synthesis system based on pre-calculated 8 bit waveforms in the early 1980s. Wavetable once meant PPG. On their collapse it passed to Waldorf, but since then pretty much everyone on the block is flogging at least one wavetable synthesiser.

Wavetables aren’t samples. A sample is a recording with many changes of pitch and phase. Wavetables can’t include this information – you have to use spectral re-synthesis. But wavetables can be re-pitched and stretched and tweaked like a pure oscillator. You can find pre-made wavetables in the Novas, the Virus, Hydrasynth hardware and others but it’s still mostly created in software.

Although processing power has grown to the point where software synthesisers like UHe Zebra can morph spectra on the fly, rendering a wavetable is still an efficient low cost way to produce a complex texture. The launch of Serum in 2014 brought wavetables into the hip hop dude-stream and the gold rush is still in effect – currently won by Vital and seriously lost in PPG’s attempt to reclaim the field with a range of odd and clunky iPad apps.

Only recently has hardware begun to directly compare with Serum. KORG’s modwave is the closest. We do not know yet how it will go.

Transwaves are a hack where a loop point is moved through a standard sound sample organised such that looped ‘frames’ have smooth joins. Ensoniq developed this technique, the most infamous example being the ill fated FIZMO synthesiser but also the TS-10, ASR-X and others. For years I believed that transwaves were somehow able to keep pitch information because the loops were larger than single waveforms. It’s not the case apart from tucking several waveforms into one cycle which allows octaves to be included. To celebrate my disillusionment I have converted a set of FIZMO transwaves to use in Vital. Sigh.

Link to wavetables from the Ensoniq FIZMO.

Geoblending is where the shape of the waveform is morphed over the duration of a sound by moving the points that draw it. One exemplar being U-He Zebra and Zebralette (and of course Vital does it too). You can’t emulate a sound sample this way but a human being can design and control noises that have similar spectral movement.

The KORG Wavestation used a later idea called wave sequencing, now advanced in the wavestate. This stitches sections of sound samples together so you can have all kinds of changes in pitch, phase … anything you like. The problem is that samples don’t join up without phase cancellations and they don’t change pitch without changing their duration, so you have to use multisamples. I have a wavestate because it doesn’t work or sound like wavetables at all. It’s the opposite of the Hydrasynth. It’s good to have creative differences.