Extracting Harmonics of Ensoniq SQ-80 Waveforms with FFTs (Fast Fourier Transforms)

[@JonDeth]

It's an interesting task relative to what I've been studying lately which is the now extinct rack mount exciters that add artificial harmonics to your signal but with a surgeon's scalpel if they're good ones. *Anyway, I'm amazed how many effects have died out with the digital takeover since digital can give us control over harmonic content like never before with current technology.* Stacking high, even order harmonics is the primary difference between "new" and "old" styles when it comes to guitar distortion. *With the valve obsession today, I'm amazed nobody has added a module based on exciter technology where you can select a large array of band points to add and remove harmonics artificial or not.* This would be the easiest pathology to make virtually any distortion including an analog pedal plugged into such a processor to sound identical to tubes, as well as produce spectrum of tones that have never otherwise existed before. *It means "your sound" can be attained from any style of distortion a modeler offers or other processing that manipulates harmonics.*

[@Lantertronics]

Pinning this (for a while at least). :)

[@edjwise]

Aaron, thank you for your dedication to the art of synths.

[@Lantertronics]

Thank you for your kind words!

[@PATRIK67KALLBACK]

Thank you Aaron for take us deeper into the FM synthesis. Even if the technology has been there for ages, its hard to wrap the math and digest it.

[@VincentZauhar]

Having an SQ-80 by my side and after watching your videos for some time now, it's very cool to see this form you. Thanks!

[@Lantertronics]

You are welcome!

[@StingrayOfficial]

Nice! Will have to watch this. I love my ESQ-1, nothing like it.

[@Lantertronics]

Thanks!

[@kwgm8578]

Aaron, I always enjoy your videos. There was no such program where a professor taught EE, CS, and music electronics in an undergrad to MS or PhD program when I went to school (1970s). Keep it up!

[@Lantertronics]

Thanks!

[@chinossynthesizer705]

Use to have a sq-80 i loved it but sometimes it would just break out of nowhere. the waveform multiple samples rips are useful for 3rd wave groove synthesis.

[@ritheshravikumar8314]

Prof.Lanterman! Great to see your take on classic synths, as always. I'm one of your "Analog Circuits for Music Synths" students from Spring 2015!

[@TheGmr140]

Good video on sounds and the fft 😊

[@Don-ev5ov]

I learned of Octave through you and have been writing simple code. Thank you. Long ago I applied the FFT in a comp-sci class to multiply polynomials (convolution) and look forward to experimenting with your code.

[@mannibimmel09]

best fourier joke. thank you)

[@Lantertronics]

I do love that one.

[@wickedprotos1937]

I wish you had been a prof when I was in College in the 80's. I was doing dig reverb out of HS.

[@lars1588]

I have a Kawai K5000S which can generate sixty-four harmonics per source (and another sixty-four from in an additional source for a total of 128 unique harmonics, or 384 if all six possible sources are used), so it would interesting to try recreating the SQ-80 waveforms on that. I also have an Ensoniq Mirage, which can be coaxed into doing wavetable synthesis with the custom Soundprocess OS, so I wonder if these original SQ-80 waveforms could be ported over to it somehow.

[@Lantertronics]

That should all be totally doable. :)

[@DonSolaris]

Thanks. Nice utility and a good reference point for the ROM waves. On the actual ESQ/SQ synthesizer above the DCA level 49-53 (depending on the wave) in pre DCF stage its harmonics start to change due to analogue circuits involved. Actually they even change from voice to voice. I don't know if they modelled this behavior in software. It would be interesting to compare.

[@stefano.a]

Perhaps it is important to precise that the FFT (the list and the amplitude of the harmonics of a signal) doesn't permit to rebuild the original sound. The phases ot that harmonics is also necessary.

[@chinossynthesizer705]

I have a Kodamo essence fm mk2. it says i can import and load 16bit, 512-sample long waveforms. If its not the right length or attributes, It will be automatically converted. Do you the thing the sq-80 wav files are to big?

[@Lantertronics]

I think a lot of the SQ-80 wav files are natively 512 samples long if I recall correctly. Anyway, if the Kodamo automatically converts waveforms with different number of samples you should be good to go anyway. ;)

[@chinossynthesizer705]

@@Lantertronics OK thanks.

[@marklewis4024]

I’m working on a project involving this. However I’m trying to understand the mapping of the waveforms.

[@Lantertronics]

You mean like how the data in the ROMs is stored?

[@SameAsAnyOtherStranger]

​@@LantertronicsYeah. Let's go with that. Prolly not gonna mean much without knowing how the D/A circuitry is producing sound, but it might be interesting.

[@SameAsAnyOtherStranger]

So...this is like FM synthesis, only with an extra frequency being modulated or is it an extra modulation wave form being applied to a frequency...if either of those? Anyway- love your work, love the channel. I kinda wish it would stay in the same lane as it seemed to be in when I first found it which was the synth module/stomp box lane, but I definitely appreciate how you are demonstrating how many lanes there are for the aspiring Electronic Musician to discover.

[@VincentZauhar]

There's a PDF/spreadsheet on Rainer Buchty's website explaining the memory layout for all multisamples zones

[@marklewis4024]

@@Lantertronics yes, I’m trying to extrapolate the individual waves. I understand the map explains which octaves the waves belong to but I don’t quite understand the addresses of each wave. If I’m correct the waves may be of various length not a fixed size. So I need to figure out the addresses for each of the waveforms.

[@bobvines00]

Aaron, this question isn't directly related to this video, but you *did* mention Octave. I believe that I read somewhere that Octave was no longer "the best" alternative to Matlab and that (I *think*) SciLab has taken its place. Am I remembering this correctly?

[@Lantertronics]

I view them as different tools for different purposes. SciLab isn't trying to be MATLAB compatible -- it's trying to be its own thing. I've used MATLAB for well over 30 now and am very familiar with it, so I naturally gravitate to Octave just for speed. SciLab is slicker in appearance in some aspects, and is probably better designed as a "language" (MATLAB sort of accreted features over time and doesn't come across as well thought out). I'd say try them both out for whatever purpose you have and see what you like better. That may vary from use case to use case.

[@Lantertronics]

My main use case for SciLab is for the open-source FPAA development tools developed by the research group of my colleague Jennifer Hasler. It uses the Xcos block diagram language, which is like SciLab's take on Simulink. It's excellent.

[@Lantertronics]

Here's my FPAA playlist, you can see a lot of SciLab and Xcos: kzbin.info/aero/PLOunECWxELQTCZEqIbHZpRmIEeQR80o3j

[@Lantertronics]

Oh! And if you are wanting something closer to MATLAB, but not necessarily MATLAB compatible, check out Julia. It's like what MATLAB would have been if had been thought out well from the beginning.

[@bobvines00]

@@Lantertronics Thanks Aaron!

[@AnalogDude_]

So if i understand right, i could store different samples for each octave on the keyboard to be a wave station synth a like?

[@marklewis4024]

@@AnalogDude_ yes, that’s actually how multi sampling works.

[@AnalogDude_]

@@marklewis4024 that how a "piano" program on a akai sampler, like S3000XL works, Waldorf wave station also?

[@marklewis4024]

@@AnalogDude_ yes… I also have an Akai Z4 and that’s typically how most samplers work. There may be a sample per every 3 keys (e.g.- C, C#, D and maybe D#) then another sample for the next group of keys.