This video, via Fabbri’s Midnight Librarians, offers an introduction to Transwave Synthesis.
Transwave Synthesis builds on simple wavetable playback by allowing you to loop waves, and then control and modulate the loop points, allowing for complex, evolving sounds.
The video follow this with a demonstration, using an Ensoniq EPS-16+ to create complex ambient pad sounds.\
Ensoniq were one of the most innovative companies back in the day. I think the first to feature this Transwave synthesis was the original EPS model, which is also a great machine.
I’d suggest to also check out The Daydream Sound, his videos are cool.
What makes this different from granular synthesis?
It’s similar but only a single grain is played instead of many
Transwaves are being used in a very interesting way in this video and in the linked videos.
The transwave modulator divides the sample up into discrete segments (probably 128 — but I don’t recall) and allows you to move the loop position pointer across those segments using any available modulation source. If you use the correct sample length and loop length it acts exactly like a wavetable. You must then make sure that each looped “segment” is perfectly in phase to all the other segments to avoid clicking and popping while jumping from one to the other during modulation. In these examples they’re trying to use a bi-directional loop to eliminate the clicking and popping within each segment. But since some segments themselves are out of phase from each other, it sounds jarring or clicks as it’s being modulated.
Normally a standard sample would need to be processed to meet these phase requirements, just like a wavetable. To fully appreciate the capabilities of transwaves you need to plan/process the sample source ahead of time.
Certainly nothing wrong with using it however you like. It’s sounds cool. But transwaves were designed to accommodate wavetable synthesis and the technology behaves just like an old PPG if sample length, sample loop length, and sample content are considered. But as soon as you go beyond single-cycle waveforms with stuff like longer loop lengths or phasing differences, you’ll get weird behaviors and sounds as the modulator moves to the next frame mid-cycle. Adding the bi-directional loop type into the mix just makes it even weirder. It’s lovely and probably why it sort of sounds like granular synthesis.
Transwaves can be looped at any point, and preceded granular synthesis by a decade.
Great overview but there’s clearly a lot more going on, your demo included some sample stretching which is non-trivial and if you simply move window loops points about you risk clicks & other artefacts so there’s clearly more going on (and it’s not granular synthesis, unless the grains are possibly 1 wave-cycle in length).
Is there an “analysis” phase when you load a new sample?
Is there more detailed information about “Transwave Synthesis” available?
Thanks!
There is not a realtime analysis phase. To avoid artifacts, one needs to be sure the wave file being used has zero crossings at the loop points.
Ensoniq, the one and only “true” sampler.
I’ve owned a 16+ for 20 years and always wanted to try this but I’ve never been able to get hold of the Transwave disks in physical form, just files I couldn’t write with a pc disk drive. Any help is appreciated.
If you are a PC user, look up tranzilon. It’s a tool for generating transwaves. The developer also has tools to transfer the output directly to the EPS 16.
Love Ensoniq. The ESQ… yes.
Nice introduction, but a few important points need to be clarified.
The example of transposing samples was a bit misleading (as mentioned by Richard Meyer, above). In traditional samplers by Ensoniq & Akai, samples were played faster/higher, or slower/lower using either a variable playback rate, or by adding or dropping samples. His example used pitch-shifting that did not affect playback speed, and had artifacts. This was not the case with the EPS or Akai samplers which had full “munshkining and vadoring” and very little of what we would call “artifacts”.
Next, the narrator refers to loop-point modulation as Transwave synthesis. Yes, the Ensoniq EPS could modulate start, end, loop-start, loop-end, loop start/end. But this is a more general feature of the EPS. A MIDI CC could generate 127 steps (128?), so the EPS had that limit to resolution when moving any sample pointer across the length of the audio sample. Still, it was an interesting feature. Transwave synthesis applies specifically to synthesis that uses a special sample that was made up of a series of single-cycle waves appended to each other– like a wave table. The loop length had to be one cycle (though a loop of two or three cycles might have worked). The EPS could then modulate the start and end points of the single-cycle loop across the sample, moving through different single waves, and creating new timbres along the way.
This differs from wavetable synthesis in that a wavetable synth cross-fades/morphs smoothly between the waveforms on the table, without having to cross through the continuous phase states to do it.
I think it is probably more similar to granular synthesis in that the loop is like the grain. But transwave synths lacked any crossfading capability (if I recall correctly) so there wasn’t any smoothing as you swept the loop through the waveform.
At times, I’ve waxed nostalgic about how useful it would be to have modulate-able sample pointers with my current samplers. Combined with some ability to smooth out the result, it could be a useful feature. The EPS’s transwave feature was probably inspired by the PPG in some respects.
you could use this method on an octatrack too and it would be dope
You can and it is.
Ensoniq EPS and ASR were fantastic samplers, unfortunately their operating system was very problematic, leading to crashes, freezes and since there was no way to download updates back then, it was very cumbersome to keep the machines up-to-date. One had to either purchase from Ensoniq US (which took a long time to get the disks) or hope that some of the few dealers could order them.
The power of the machines themselves were very limited so even small offline sample manipulation took a lot of time with sometimes unpredictable results. Nevertheless, I really miss my ASR10-R …
Yea. It was the most unreliable major piece of gear I owned. It made some incredible sounds sometimes during a crash.
I’m not sure if it was the operating system or whether it was a hardware/noise issue that caused things to glitch.
I’m with you though. The EPS 16+ and/or ASR10-R are a couple things that I’d like to see modern takes of, where the obvious shortcomings are addressed and brought up to date.
Huh, time to dust off the old EPS 16+ again. Although, I had a Fizmo and it was too cheap feeling.
The EPS and EPS 16+ does have polyphonic aftertouch, too!!
Transwave synthesis is exactly the same thing as the original wavetable synthesis in the PPG Wave synths. They didn’t crossfade between the different waves in the wavetable either, which is evident if you sweep slowly through a wavetable on a PPG Wave 2 or any of the emulations out there; the resulting sound is very obviously stepped. More recent implementations of wavetable synthesis crossfades between waves, but the old PPG synths didn’t have the computational horsepower to do it, with just a humble Z80 microprocessor to handle everything.
That’s really interesting. I didn’t know that about the PPG. I had always assumed it was crossfading running oscillators. But yea, now that I type it, that wouldn’t make sense from that year. Still, it must have somehow gotten more than 128 steps.
It is a shame that Ensoniq did not implement their ‘Hyperwave’ technology from the TS-10/12 in to a sampler, this is where two Transwaves can be cross-faded together to produce a smooth morph. It is also a shame that the implementation with the TS only allows for the onboard Transwaves to be used and not those of user creation.
The Ensoniq EPS-16 Plus likes single cycles of 256 samples…so…to make a sample that can be transformed into a Transwave it is best to sample the F3 key tuned to -23…this will get you a sample that consists of single cycles of 256 samples.
Then set the looplength to 256 samples and set the MOD to TRANSWAVE and SRC to MODWHEEL. You can now scroll through the sample with the MODWHEEL.
What I did in the LINKED TO video (Method 2)though is to sample 8 sounds for 2 seconds each, pasted them one after the other so I would get a 16 second sample. I looped each segment using Loop Bi-direction. Then I used one of the ENVs in REPEAT MODE to modulate (scan) through the 16 second sample.
Using other modulators as LFO, Filter Env and Effects creates an (ever) evolving soundscape.
Can this be done on an original EPS?
Transwaves were introduced by Ensoniq when they released the successor of the EPS, the EPS-16 Plus, so this cannot be done on the EPS.
But, I think the EPS does support the modulation of the loop position…so you could play around with that and see what you come up with.
Can Halion 6 or the latest Kontakt do anything similar?