Gameboy noise channel

[Cerrax]

EDIT:I've made a soundfont of this for those interested.

Right-click --> Save As

http://www.cerrax.com/GameboyNoise.sf2

Its just a 8-bit and 16-bit LFSR mapped onto a keyboard. Toss a bitcrusher on it to get the correct sample rate.

[analoq]

I'm using this as a reference: http://www.devrs.com/gb/files/hosted/GBSOUND.txt

It seems the two modes are switching between an 8-bit and 16-bit LFSR so you're not exactly going to be able to duplicate that in a regular softsynth. I'm going entirely on theory here but if you took a white noise sample into a sampler and reduced the loop cycle point to a very small amount you may get a similar sound.

If you could post some sound examples of the 7 vs. 15 step cycles that might give me some more ideas.

[Cerrax]

I'm using this as a reference: http://www.devrs.com/gb/files/hosted/GBSOUND.txt

It seems the two modes are switching between an 8-bit and 16-bit LFSR so you're not exactly going to be able to duplicate that in a regular softsynth. I'm going entirely on theory here but if you took a white noise sample into a sampler and reduced the loop cycle point to a very small amount you may get a similar sound.

If you could post some sound examples of the 7 vs. 15 step cycles that might give me some more ideas.

I've been using this as sort of a guide: http://www.belogic.com/gba/channel4.shtml

I don't really have any audio examples, but if you've ever messed with the Gameboy Camera and played the little DJ game Trippy H, the 15 step cycle is Noise 1 and the 7 step cycle is Noise 2 in the Noise tab. Like I said the 15 step is very close to real white noise, but the 7 step has a more metallic, harsh tone (since there are less steps to define the wave).

Yeah I've been experimenting with a 16 bit LFSR and doing just as you say by gradually changing the loop cycle and the sample rate and I'm getting mixed results at best. The algorithm for the Gameboy noise channel is more intricate than that.

With the amount of time I've already invested researching the subject, it seems that trying to recreate an inferior noise generator with superior hardware is just going against the grain of productivity. There are a lot of factors working against this, many involving the hardware on both ends of the process. So I guess the Gameboy wins. If I can get a worthy soundfont together, I will be happy to share it with all who long to hear that classic noise.

[analoq]

If I could be arsed to hunt down examples myself then I wouldn't have asked for them. I've never owned a gameboy of any sort.

Yeah I've been experimenting with a 16 bit LFSR

Huh? How so?

doing just as you say by gradually changing the loop cycle and the sample rate and I'm getting mixed results at best. The algorithm for the Gameboy noise channel is more intricate than that.

It's not any more intricate than that, the problem is finding a grain similar to the psuedo-random pattern is apparently a crapshoot.

it seems that trying to recreate an inferior noise generator with superior hardware is just going against the grain of productivity. There are a lot of factors working against this, many involving the hardware on both ends of the process.

You're making it out to be a bigger problem than it is, there are no hardware limitations at work here. Writing an 8-bit LFSR noise generator and wrapping it in the VSTi framework is not difficult, but unless you're a programmer that's probably not an option.

When you're sampling channel 4 just try to isolate the cycle. And playing it back at faster or slower pitches should approximate clock divider changes.

[Cerrax]

If I could be arsed to hunt down examples myself then I wouldn't have asked for them. I've never owned a gameboy of any sort.

Huh? How so?

I downloaded a noise signal generator program with an adjustable loop cycle and I'm using a bitcrusher to drop the sample rate.

You're making it out to be a bigger problem than it is, there are no hardware limitations at work here. Writing an 8-bit LFSR noise generator and wrapping it in the VSTi framework is not difficult, but unless you're a programmer that's probably not an option.

When you're sampling channel 4 just try to isolate the cycle. And playing it back at faster or slower pitches should approximate clock divider changes.

Well I'm not a programmer so making a VSTi isn't an option plus Logic doesn't use VSTi. I'm going to experiment some more with it, but it may come down to just sampling the actual noise.

EDT: Well I figured it out and made a soundfont that is very similar, though not exact. Thanks for your help.

Right-click --> Save As

http://www.cerrax.com/GameboyNoise.sf2

Its just a 8-bit and 16-bit LFSR mapped onto a keyboard. Toss a bitcrusher on it to get the correct sample rate.

Thanks again.

[analoq]

Well I'm not a programmer so making a VSTi isn't an option plus Logic doesn't use VSTi. I'm going to experiment some more with it, but it may come down to just sampling the actual noise.

I use Logic, what I said applies to Audio Units as well. You didn't mention what host you were on so I appealed to majority.

Glad you're making progress, though.

cheers.