I find it hard to see 2007 as that long ago, but it’s just after the dinosaurs as far as music technology is concerned. Put it this way – your newborn will soon down beers and vote. 17 years ago a small French company tried to interest the music world in a new ‘frame synthesis’ and got the usual brush off. Here they are again after a considerable pause, with some new code inside the old interface.
What is frame synthesis? They tell you in the manual how to use it, but it’s still not clearly explained. It’s not wavetables, which first came to the Average Enthusiast in 2008’s Blofeld. It’s certainly not additive either, which still needed a Synclavier or at least a K5000 at that time. I think it’s closer to Ensoniq’s transwaves. When supplied a sound sample the software will chug away breaking up the sound into sequential ‘frames’ that constitute a ‘movie’. I would guess that ‘frames’ take place on zero crossings, which like transwaves could allow changes of pitch not possible with wavetables. Each frame is a complex waveform (not spectra) that can be redrawn with a pencil or remapped with overall transforms, such as PWM or mirroring. The most important transforms are those that attempt to correct glitching phase issues between the frames.
Like Ensoniq transwaves the sound can be buzzy, increasingly so as the pitch is raised. Mellow sources are better. It’s in no way comparable to the state of the art like PadShop and so you’ll probably use it as synthesis rather than a sampler. It’s raw, you’ll try use the filter to tame it, but that’s not quite up to the job. Most of the drawing tools supplied make pointy changes to individual wave frames and even more buzz. Like the company says “Unlike modern FFT analysis, F.A.T will use waveforms to create the oscillator, resulting its specific sounding.” Just call it retro, everybody happy.
The best case for drawing is to work on one frame alone and draw the waveform gently (and in real time while playing notes if you like). You can run a ‘movie’ as you draw and dynamically carve away parts of the waveform like a spinning lathe but there doesn’t appear to be a way to paint on all of them at once. As far as I can see there’s not a direct way to create two blank frames for more complex morph effects, but you can load up a very short ‘movie’, erase the frames and draw morphs that way. That works well.
There’s also quite an array of LFOs – five of them per oscillator and able to cross talk. Very easy to apply with a drag and drop to the parameter. Along with the standard LFO waveforms you can draw your own shapes. So while a triangle LFO will play forward and back, your drawing curve might linger here and there in the ‘movie’. An LFO can also act as a small sequencer. Modulation is generally a high point in KrishnaSynth and useful even if you don’t use frames at all.
The interface is old school but perfectly usable. Unfortunately there are some mouse operations that can lead to a crash – the code is not quite restored yet. The file system is a bit clunky. The FX are OK but of a different time. Definitely get the free demo and see if it gels with you, but as you can’t import your own ‘movies’ you can’t really tell if it will work with your own samples.
I personally wouldn’t use this as a workhorse, more likely I’d play with it to get a start on something unusual and move on to more reliable tools.
Krishna is the only available transwave capable VSTi, at least it was a few years ago. In preprocessing the input audio, it’ll somehow homogenize each single cycle wave so that the zero crossings occur precisely at the beginning and end of the so called frame. Since all frames are identical in length, they can be mixed and matched via modulation, at will, cleanly. This isn’t really doable in a wavetable category synth. Anyway I stopped investigating before I reached a logical conclusion. They re-released it what last year or the year before. This one is better, more stable, I believe and there’s a 64 bit one. I really need to revisit.
Sorry for spamming a second entry – I just re-read there are 6 LFO’s of hand drawable shape available. Assigning one or more (added together perhaps?) to frame read position, would show off the transwave concept in the fullest way. I guess.