NeverEngine Labs™ is proud to present
Additive Modal Analysis and Synthesis for Kyma 7
SciPhy Modal Analysis
The SciPhy process begins by extracting modes of an instrument or timbre which has been previously analysed using Kyma’s Spectral Analysis tool to create a Kyma SPC file. The modes (or partials) can be thought of as the resonances of an object being excited. For example a perfect string would contain only integer multiple frequencies of its fundamental (and is therefore perfectly harmonic), while the modes of a metal plate show a more complex, non-harmonic behaviour.
Conventional Mathematical models calculate modes using formulas which describe how soundwaves behave in the real world, but we didn’t go that way. Instead, using our ModalAnalysis prototype, you are able to extract those modes from any audio recording of any instrument or sound!
The SciPhy doesn’t output audio data – it’s a Spectral Source – it outputs Kyma’s standard format of a frame based stream of Amps on the left channel, and Freqs on the right. To hear the results of the SciPhy, you must resynthesise at some point in your signal flow using one of Kyma’s Additive Synthesis classes (OscillatorBank, FormantBank,…) or one of our NeverEngine Labs StackOscillators. This process transforms the modes into partials which you can now manipulate with envelopes and shaping. Right, so the conventional ADSR concept is pretty straightforward. But the interesting part in SciPhy resynthesis is setting the responses of each partial to the ADSR parameters – individually!
Each partial gets its own individual ADSR envelope and LFO modulation and can respond to it differently.
It would be a time consuming process to set 16 ADSRs and 32 LFOs (16 for AM, 16 for FM) by hand, so we made that accessible in an algorithmic and meaningful way. When the AttackResponse control is set to a positive value the attack time gets shorter for higher frequency partials. This resembles metal objects where the speed of sound is frequency-dependent and therefore higher frequencies travel through the medium faster. When setting it to a negative value it’s the other way around, lower frequencies will have a shorter attack time then higher frequencies. This concept of using a single response control to adjust settings for all modes individually is followed at many points. The DecayResponse and ReleaseResponse work the same way. The SustainResponse works very similar but here we are dealing with amplitude not time. So when setting it to a positive value the amplitude gets lower the higher the frequency, and when using a negative value it’s the other way around.
The Modulation possibilites of the SciPhy are quite exceptional. First of all you can smoothly morph between 141 waveforms using the AM/FM Shape. The currently selected waveform is displayed graphically. There are 3 types of response controls for the amplitudes and frequencies modulations:
While the SciPhy produces interesting sounds on its own it does lack having a body, or so to say it doesn’t contain any kind of reverb. A signal without reverb is naturally impossible, there’s no sound without space. So we designed this kind of reverb object especially for the SciPhy. Using ShadowObject and Dry you can adjust the mix between the SciPhy and the Reverb. If you think of it in physical modelling terms it’s like attaching the sound generator (the SciPhy) to a resonating plate that changes its shape according to its input. The length of the plate contains odd harmonics while the width contains only the even harmonics. As well as ShadowObject we have also designed an example of how you could resynthesise your partials manually by replicating your own bank of oscillators, and why not then add a delay line to each partial, with variations derived from the modal analysis information? This is the StackDelay effect. Although this is not strictly the conventional ‘waveguide’ technique from the physical modelling literature, it does introduce varied delays as a parallel network which can smear and warp in the SciPhy resynthesis in very interesting ways, and with a very high quality.
Design & Programming by Gustav Scholda and Cristian Vogel
Concept & Idea by Gustav Scholda
Artwork & Wavetables by Cristian Vogel
© NeverEngine Labs 2016
Compatible with Paca and Pacarana hardware. Requires Kyma 7.
SciPhy basic gong no effects
2 Glass taps then 2 taps with Shadow Object Effect
Vogel 90s Preset
Vogel BlueBendir Preset with and without Slack Delays Model A
Noisy Attack with Stack Delays Model F
Gustav Scholda created this short demo of some of the presets
… The SciPhy is built on research from the Spectral Lab … Sign up to discover even more additive and spectral sound processing techniques!