Manatee - From concept to instrument

[marzacdev]

Dear friends and users at Modwiggler.

In May 2022 @ Superbooth, I presented the Fred's Lab Manatee prototype, or better put, its proof of concept

If you have not heard about this project, here are some presentation videos:




The Manatee is my personal attempt to offer a purely digital machine with a quite distinct character, moving away from the overdone / overheard Virtual Analog architecture and embracing the special functions a DSP can offer to sculpt the signal, rather than trying to emulate or reproduce anything in particular.

The actual sound engine was made from scratch. It has undergone several revisions and paradigm shifts and started as a research project on spectral synthesis. It is far from finished and the spectrum manipulation tools need more research and try/error/correct cycles, to end up with the best (musically speaking) algorithms that can run on the selected platform. For those interested in the process, once formalized, I transpose my DSP ideas into simple C algorithms, run, debug and tweak them in a custom audio framework. Finally I integrate them in the signal chain ensemble. Once the blocks seem to stabilize (well, when I am happy with them and no major change is planned), I start to think about how to properly arrange data and computation, in order to write clean and efficient DSP code.

FYI, the Manatee audio engine is mostly coded in DSP assembly, for obvious reasons of performance and to take advantage of the dedicated signal processing resources.

As done previously with the Töörö project (in another community), I decided to open this thread as my dev.blog.

I will share with you the current state of affairs, my doubts and all experienced difficulties working on this machine and listen to constructive criticism and comments. I will ask you to go easy on me, developing machines like these is a huge work, a real marathon for a single developer, and many things just cannot be done, and for other than purely technical reasons. Time, money, assets and company business strategy are also involved therefore do not expect me to justify myself on certain design decisions that may look odd to you.

Off we go!

[unclebastard]

I like the sounds and the concept: the only other spectral synthesiser I've used is the Virtual ANS, which is weird, but not easy to create musical sounds with. It excels at otherworldly stuff, but because the microtonal bands are so close together defining pitch is difficult.

[marzacdev]

Hi unclebastard,

I have never heard about this one, thank you for pointing it out to me, I will check its features.

The Manatee is mainly intended to produce soundscapes, pads, chords...
but it can be used as a regular VA, chose a single harmonic spectrum, a basic reconstruction waveform and no spectral animation and here you go.
Current sound engine handles very well fast modulations, as it is not modular internally, so it's great for percussive sounds and crazy effects!

Now, I am redesigning the user interface and housing form factor.
Since the machine uses pots and not encoders, I am thinking about having a second row of pots for the 2nd envelope. This has been suggested at the Superbooth and it actually makes the editing easier. Also more control on the spectral section as this is the main attraction of this thing.

The panel could also be made so that the machine can be 19" racked, not everyone (including myself) being fan of the small desktop format.
I am looking forward to hear your opinions on this too.

[Faru]

this sounds great so far!
my initial thoughts:
I have never been a fan of menu diving and this probably ties into the domain of cost, but the best instruments ux are the one where dont need many menus (ms20, moog, etc) or non at all ideally. Personally I also dislike having same knobs for different function, this always put me off the walddorf instruments, because you can never see which state its in when u change the menu - but this is hard to judge without using it. but other people love it, so hard to say. If you want to do some proper research, I would recommend doing a ux test, define some tasks and see where people struggle. Opinions are just that and if u go with many opinions you end up with a homer car - instead of finding out what need you want to serve and problem you want to solve. (id be happy to give some more info if you are interested)

The shruti for some reason I cant put in words, manages to be a lot of fun, even tho it has a lot of menu diving. but again, lots of people dont mind and its probably a question if you aim to serve sound designers who like to fiddle around or rather something fun for performance and jamming.

if you could post some more detailed photos of the interface, it might be easier to give some feedback. I think the new elektron syntakt is a great example for UI design limited in color, space and resolution and putting in meaningful and joyful icons to enhance the ux.

I like the aesthetics of the case so far, kind of industrial / construction site themed to me and reminds me a bit of the wmd pedals.

[marzacdev]

With the current firmware, except for configuring MIDI channels and such ...
the Manatee has simply no menu diving.

There are a few modules that attributes multiple functions to the same pots, there is no way around it.
The UI is / will be better arranged to allow editing of the sound basics without having to change pots attribution too often.
This will also come in handy when performing with the synth, you don't want to dig out your access to a critical parameter when performing.

Here are some highres. pictures of the machine:
https://www.synthanatomy.com/2022/05/su ... sizer.html

[unclebastard]

This is the Virtual ANS:
it's an interesting synthesiser- the original electromechanical one weighed two tons and took twenty years to build- Coil made an album with it. https://warmplace.ru/soft/ans/

[Faru]

With the current firmware, except for configuring MIDI channels and such ...
the Manatee has simply no menu diving.

There are a few modules that attributes multiple functions to the same pots, there is no way around it.
The UI is / will be better arranged to allow editing of the sound basics without having to change pots attribution too often.
This will also come in handy when performing with the synth, you don't want to dig out your access to a critical parameter when performing.

Here are some highres. pictures of the machine:
https://www.synthanatomy.com/2022/05/su ... sizer.html

nice, thats great to hear!

[budz]

looks/sounds intriguing.. I would be very tempted if the finished thing supported MPE and had outputs per voice, would love to post process the poly voices in a modular. Good luck with the development.

[marzacdev]

The Manatee will support a reasonable subset of the MPE specification.
You will probably be able to select MPE in mono-timbral mode or the regular multi-timbral mode but not both at the same time.

Individual outputs per voice is out of the question, it's way too expensive and cumbersome to do on this architecture.
There are already two line buses available.

Thanks for the kind words.