Well I thought I had the perfect set up with braids and rings, now I am not so sure. Great stuff where the interface is concerned, I am looking forward to trying out Plaits as a single voice synth. Maybe I should sell one of my rings modules and pick this up? Too bad the ctrl-mod pre-order is sold out.
I had a wee bit of time this afternoon and a few folks had shown some interest so I put together a little exposition focussing on the talky words side of Plaits.
nice one! but you have to move the camera a little bit more to the left
Thanks for making that. The Mutable Instruments Marketing Department missed a cheesy trick by failing to have the names of all the Mutable modules included as a word bank (including the names of some yet-to-be-released modules). It would have rivalled Elon Musk’s recent cross-promotion of his forthcoming Tesla sports car…
Meanwhile, the Department of Marketting Dept people over at Synthesis Technologies must be twitching as they listen to your nice demo of this feature.
Anyway, the vocabulary will be able to be adjusted when the source code is released. First thing that needs to be done from my perspective is to replace that horribly mispronounced final letter of the English alphabet: it’s zed, not zee. I mean, did Peter Greenaway ever make a film called “Zee and Two Zeroes”? No, he didn’t!
first impression didn’t like to much but after watch the Divkid video , I change my mind.
I want one , good job !!
Looks like the Learning Modular channel will be making a series of videos as well. This is one of those modules where I feel like I will be learning something new.
Algorithms for generating waveforms (like sawtooth, square…) that have very little aliasing, especially when played at high frequencies.
Instead of synthesizing a naive sawtooth (and dealing with the aliasing it can introduce), you synthesize a waveform that has the same spectrum as a sawtooth in the audible range (up to 20kHz), but no other harmonic outside of this frequency band. And thus, no risk of aliasing. Your ears won’t perceive any difference, and on an oscilloscope it shows as slightly curved edges on your squares and saws.
Do you use Poly Blep for all the waveforms, as well as the LPC synthesis, then?
Ok I thought I was over needing any new modules but this one I can’t resist. It looks totally amazing. Incredible work Olivier.
Need to raid the piggy bank. Should I expect these to sell out quickly? Will there be a delay between production batches?
Yes, for all models that have discontinuities (VA, string/organ machine chords, variable rate LPC, granular cloud, clocked noise, cymbal noise).
- For FM: 4x oversampling + FIR filter.
- For wavetables: integrated wavetable synthesis
- For waveshaping: rough estimatation of the spectrum shape, then “taming” of the amount of waveshaping/wavefolding as a function of the fundamental frequency (like on Tides).
Most dealers have already sold modules from the first batch shipped to them. The second batch is at the end of May. I’ve launched in parallel a new batch at a different manufacturer, maybe they’ll make them faster!
I got the Braids few months ago, I still love it. I was wondering if Plaits could be a good complement.
I won’t have the chance to test it so your answer will help me
Thank you !
I’m super interested to know how the speech synthesis noises are defined, is this something that could be compiled with ones own “dictionary”? That would be a lot of fun. Now I say it out loud, sounds like more of a job for a disting or something with an sd card…
Wrt future firmware hacking, I wonder which version of the STM32F373 Plaits uses? In particular, does it use the version with 128k of FLASH storage, or the one with 256K of FLASH (and a bit more RAM)? And I wonder how much of that storage is currently occupied by the factory firmware?
I’m not so much thinking in terms of new oscillator/DSP models (designing that sort of thing is above my pay grade, although the maths is certainly interesting, and the more “hard-coded” UI of Plaits doesn’t lend itself quite so well to adding or substituting oscillator models, at least not without a new faceplate). I’m thinking more in terms of space for data of various types…
Presumably it has FTDI serial and JTAG headers on the back? Maybe someone could post some photos of the rear of the module? SchneidersLaden doesn’t have it’s usual side and rear photos of the module yet.
No dictionary. This is more like playing extremely compressed (as in a few hundreds byte / second) samples. There’s an old Windows 3.1 program around (QBOX Pro) to do the conversion from .wav files. https://www.ninerpedia.org/wiki/Development_resources
The module is programmed by SWD, and there are no connectors on the back except for power. I now design and build a couple of test jigs (with pogo pins) for programming/testing. It saves the cost of soldering connectors, the time and occasional mistakes when plugging-in programming cables, and the occasional repairs caused by damaged connectors. Overall a few euros saved per module.
I guess you’ll have to solder a header on the back and build a little adapter board.
Cool, 256k of FLASH storage.
That’s a very sensible approach, and apart from saving a few euros per module, it may help to inhibit, at least a little bit, (um, how can I put it?) “large-scale, non-value-added advantage taking of the OS licensing model” behaviour.
Yes, although presumably the audio boot loader is still there?
Oh Boy. I’m going to need more than one of these… lol Nice Work…
Ah, does that mean a) there is provision for a programming header on the PCB layout but no physical header pins are installed, or b) that there is no provision for a programming interface except via a bed-of-nails (and of course the audio bootloader for firmware updates)?
If you have a Mac, there’s also BlueWizard (look it up on GitHub).