8 bits with a G2 , how to ?

[zgogor]

hi all,
I'm zgogor, brand new here.
I've bought a NordModular G2.
Can someone point me on a website or something to create some 8 bits ( even 4 bits !) sounds, I need some GameBoy sounds and Bits crushed stuffs to do with my G2

thank you
zGo

[Kassen]

Hi, zgogor, Welcome to E-M!

As a moderator I moved your topic to the specialised G2 area where I hope people will be able to help you better.

For a start you might considder using the "digitiser" module. From there on the question becomes what speciffic model of the gameboy you'd like to emulate.

[blue hell]

Digitizer yes, and the AD DA modules can be used of course for more flexibility.

[Kassen]

For 4 bits it might even be possible to roll your own DA and AD converter "building blocks". I'm not realy in favour of the standard AD and DA modules. Those modules count up from minus the maximum aplitude to plus the maximum amplitude. For waveshaping applications I think it's much more convenient to use one bit for polarity, then use the rest to indicate the distance between the sampled value and zero. This makes symetrical waveshaping without headaches much easier.

A large part of the sound of the various gameboys lies in the different anti aliassing filters though. The oldschool grey version sounds to my ear like it does it the oldfashionend brutal way using a analogue LP filter. I stronly suspect some amount of overdrive is involved between the DA and the filter too. Later models lack that brutal overdriven sound and the newest have probably gone with FIR filters and sound much brighter, clearer and cleaner.

[Stanley Pain]

whenever i fear, or aspire to be a geek i like to pop into electro-music. either way, it's re-assuring and exciting to realise i have a looong way to go. god bless you kassen.

i have to say, i really dig the sound on the new gameboy DS. the stereo image they have achieved is nothing short of stunning considering it's size and cost. imhhho

[zgogor]

thank you !

zgogor

[mattBlack]

try this resource (especially this faq linked from it) for more info on the building blocks of the gameboy's sound; there's much more to it than just digitization!

here's the headlines:

- 4 'semi' analog audio channels;
-- 2x pulse waves (duty width mod, one with frequency 'sweep')
-- 1x sample-based (using 4-bit DAC, up to 32 samples) or triangle wave
-- 1x noise generator & volume 'envelope'
- LPF (& HPF on the colour model)
- different models & model revisions have other variations in architecture

there's quite a big gameboy sound scene out there; to the extent of people having specifically extracted the code used to generate the tunes / fx and written emulating software to play them. it may even be possible to derive exactly how a specific sound was produced / obtain the PCM data, if you're prepared to get your geek on & wade through loads of technical stuff...

...but i'd favour just sticking the blocks together & playing with the controls myself! can't do the sample-stuff with the g2 alone though...

and my lass called me a geek tonight... which hurt!

[jksuperstar]

If the "sample-based" oscillator is only 32 samples long, you can chain 2 sequencer modules together, then pipee them through the appropriate DAC or digitizer module of your liking.

[cebec]

[Fozzie]

[Fozzie]

Wouldn't this suffice? I just patched it in my coffee break, without G2 to check, and on a slightly older editor version (but that shouldn't matter)

edited twice: thought I could make my patch more elegant, but I made an error. There was a ADDA II pch2 up here shortly that was incorrect, this is the original again, that should work properly.

---

Symmetrical ADDA.pch2
Description:	Ad/da conversion that should be capable of doing symmetrical shaping of negative and positive waveparts. Not checked for functionality	Download
Filename:	Symmetrical ADDA.pch2
Filesize:	1.03 KB
Downloaded:	1234 Time(s)

---

[Kassen]

Cebec, you aren't accidentally a registered Tassman user? This one does exactly that;
http://community.applied-acoustics.com/php/community/view_synth.php?id=87

I could do a quick building block, I suppose, but the issue is that doing proper two's complement requireres loads of logic and compare modules. It's very cheap on the CPU but it eats zero page for breakfast. Fozzie's strategy of inverting all negitive voltages, then using the off the shelf module AD for everything but the most signifficant bit might be best.

THe general stategy (aside from polarity) is compare values. You compare your input value to 32. If it's higher you output a positive bit to that output and substract 32 from it. Then (in either case) you compare it against 16 and do the same. Etc... Keep halving the value untill you have as many bits as you want.

It's not worth doing this manually on the G2 unless you are interested in introducing "errors" in these tresshold values which in the G2 would be most efficient to do at the DA stage.

Let's asume we have the polarity sorted out and are getting all remaining bits from the stock AD converter. To reconstruct the signal we take 7 VCA's and as many constants. The first multiplies the second most signifficant bit with 32, then the one after with 16 and so forth. If you mix all of those VCA together, then apply polarity (by multiplying with -1 if the orignial wave wasn't more then or equal to 0) you get a straight signal back that should sound like a more lofi version of the input. What you can do now is take those constant modules (the ones that go 32, 16, 8, 4, 2, 1, .5 blah blah blah) and set them to different values. Stick roughly between .75 and 1.5 times the value they had. This results in a kind of waveshaping that can sound reminicant of cheap DAC's. With some filtering you can get fairly close to that Amiga sound. This is a interesting and subtle technique for a extra kind of "edge".

Just doing this at the DA stage doesn't need to break the zero page bank; AD is much, much more expensive.

To get a proper "fat" Amiga/gameboy/hipster retro computing style sound with any amount of credibility on the G2 you are going to HAVE to add a tilting filter or saturation or other Rob contraption. This stuff is basically begging the DAC's to show themselves from their worst side. I don't want to go over all that again but the big advantage of my Tassman version above is that Tassman of itself doesn't alias and so all artefacts are those intentionally introduced. The usual "the G2 sounds clean" crowd may ignore this, then they will sound like 90's IDM instead of hip micromusic.

Don't forget to wear thiftstore Tshirts and a courier bag; japanese designs prefered.
:¬p

[Fozzie]

Nice idea, doing the DA with 'custom values'. You could also morph them or change them separately with all kinds of control signals. Sounds like good fun.

[cebec]

yeah, it does sound like fun! the changing the constant values part, i mean... thanks for that explanation.

[Kassen]

It' s a technique I developed myself. I stumbeld uppon some more (tangentally) related stuff. I think that what's currently going on in micromusic is realy only the tip of the iceberg; you can get much more exotic....

Happy hacking.

:¬)

[jksuperstar]

If you just want to introduce some non-linearity into a "custom DAC", you can always use wave shape modules before the DAC module.

For a 4-bit custom DAC, use a SeqCtr module as a look-up table, and turn off the X-Fade. Just increment each step by 8 for normal linear operation. With this you can tweak each DAC's output value, control bipolar vs. unipolar operation, etc.

[cebec]

[mosc]

[Kassen]

I dunno. I don't think this is any less valid then a Tubescreamer. It's just another type of harmonic distortion. If you start moving the AD treshold around as well (which my own effect I have on my laptop does) you can do some types of distortion that'd be hard to do otherwise.

Either way; you don't have to use it, Zgogor wants a gameboy style sound so for him this stuff might be wothwhile. I suppose some builders of brass instruments that were aiming for a pure tone as well would be similarly displeased with emulations that also use breath noise. It's no my fault that it works that way.

When it comes down to it breath noises can sound good and so can crappy DAC's. The nice thing about this idea is that as long as you don't clock it it's all harmonic which can't realy be said for the gameboy. Humm, hiss and crosstalk can be musical too but are hard to use unless you have well controlled humm, hiss and crosstalk. To me it's not about emulating a Gameboy or a Moog, to me it's about taking the bits that can be unexpectedly musical, isolating and distiling those and using them on their own. Don't expect me to haul around a Moog or stare at a Gameboy screen for ages.

[jksuperstar]

Also, I find asymmetrical distortion the most ugly-yet-still-pleasing to the ears. (Asymmetrical has different non-linearity for negative vs. positive values). This is most likely to be found in a DAC on a cheapo gameboy or similar, due to lack of high current supply and/or good grounding (or if the device uses a bipolar supply, the + and - supplies aren't truely equal).

Some of the the G2 waveshaper modules have a switch for asymmetrical vs. symmetrical, but they only allow for saturation near the positive end of the curve. Hmmmm, yet another application for a 256-entry lookup table module

[cebec]

[jksuperstar]

Are you reducing the bitwidth down to 8 or 4bits? If so, the saturation from the waveshaper *might* get lossed due quantization. Maybe in this situation, the waveshaper should be placed *after* the 8 or 4bit DAC

Those waveshapers are simple math functions with no "history" built in, so they are still just remapping certain values to be non-linear.

[Afro88]

Haven't read the full explanations yet, but I want to say that Fozzie's patch, when you connect up a few random outputs from the DA to the AD, is really cool!

Especially when you put a filter and a few env's in there for pitch and cutoff Great sounds!

[ian-s]

here is a lo bit DIY A2D and D2A

The D2A is just a mixer, summing and scaling the 3 output bits, changing the mixer levels will introduce errors

The A2D uses compare level modules and the subtracting mixer, one pair for each bit. This example uses the spare mixer inputs to allow some bits to interfere with other sections. Just as an experiment, the LSB has a random trigger on it, probability modulated by the middle bit.

I'm just processing a pos only signal then converting to bipolar later.

Sounds more interesting on complex signals.

---

ad3da.pch2
Description:	DIY A2D & D2A	Download
Filename:	ad3da.pch2
Filesize:	1.81 KB
Downloaded:	1450 Time(s)

---

[Kassen]