Bits

Shifrin · 7 Jun 2005

Nu Audio Science wrote: How do i gain acces to bit ranges and stuff in the code component so i can mess with bitcrushing ?

I've not tried it myself but I'd guess you need to quantise the floats to the bit range by doing something like :-

monoin in; monoin bits; monoout out;

float tmp;

tmp = in * 2^bits; tmp = rndint(tmp); out = tmp / 2^bits;

Now there isn't a 'pow(x,y)' or '^' function to raise 2 to the power of the number of bits so you can do that outside of the code component, if you wanted 4bit output you'd get :-

tmp = in * 16; tmp = rndint(tmp); out = tmp / 16;

16 coming from 2^4. There could well be an easier way to do it though I just got up 😉

Shif.

Shifrin · 7 Jun 2005

The 2^bits might need to be (2^bits - 1) so the 4bit example would be (2^4 - 1) i.e. 15. Not sure, my brain hasn't kicked in gear yet! I'm sure Unk will put me right if I got it wrong 😉

Shif.

Nu Audio Science · 7 Jun 2005

Cheers Shif i will mess around and see what comes out 😉

Nu Audio Science · 9 Jun 2005

Can anyone point me to a page explaining what this math is actually doing ?

Shifrin · 9 Jun 2005

I'm not sure my explanation will be any good, but here goes :-

Lets say we have some numbers in the range 0.0 to 1.0, and we want to bit-crunch down to 4bit. With 4bits we have 2^4 i.e. 16 possible numbers we can represent, 0 to 15.

So the first step is to multiply our number, e.g. 0.3456 by 15, this gives 5.184. Then we round this to the nearest integer, which would be 5. Finally dividing through again by 15, to get 0.3333. This process forces the output to be one of only 16 values, i.e. 0/15, 1/15, 2/15, 3/15 etc... to 15/15 i.e. 0.0 to 1.0. Does that makes sense?!

Oh yeah, I think the code I wrote above is actually incorrect, I'll have a look at it and check it out!

Shif.

Nu Audio Science · 9 Jun 2005

Shifrin wrote: Lets say we have some numbers in the range 0.0 to 1.0, and we want to bit-crunch down to 4bit. With 4bits we have 2^4 i.e. 16 possible numbers we can represent, 0 to 15.

Shif.

Thats the bit im losing it at, If its 4 bits why 2^4 ?

Shifrin · 9 Jun 2005

Nu Audio Science wrote: Thats the bit im losing it at, If its 4 bits why 2^4 ?

A 4-bit number is as it's name suggests made up of 4 bits. Each of these bits is either on or off, a 1 or a 0. Some examples would be :-

0000 0100 0011 1010

Etc, if you look at all the permutations of the above examples you'll find there are 16 in of them in total! You know in maths at school how you had units, tens, hundres, thousands etc. Well in binary you have 1s, 2s, 4s, 8s, 16s etc.

For example the 4-bit binary number 1010 means :-

8s 4s 2s 1s 1 0 1 0

Which gives 8x1 + 4x0 + 2x1 + 1x0 = 8 + 0 + 2 + 0 = 10 🙂

Any clearer now 🙂

Shif.

Nu Audio Science · 9 Jun 2005

Give me a little while to mull it over 🙂

skOre · 9 Jun 2005

Ah, the in-your-face beauty of binary 😀

GaryG · 10 Jun 2005

nicked from thinkgeek.com:

"there are only 10 types of people in the world, those who understand binary and those who don't..."

🙂

skOre · 10 Jun 2005

Yeah, thinkgeek rocks, as this sentence does 🙂

Sk · 10 Jun 2005

Isn't the effect you're describing a simple quantizer?

Nu Audio Science · 10 Jun 2005

If it is quantizing that isn't what i'm after anyway ?

I could do quantizing in SE and it sounds shite, I'm after the sound my Amiga used to add to samples 8 Bit Octamed style 🙂

platinumears · 10 Jun 2005

what sample rate did the amiga use?

Nu Audio Science · 10 Jun 2005

Nu Audio Science wrote: If it is quantizing that isn't what i'm after anyway ?

I could do quantizing in SE and it sounds shite, I'm after the sound my Amiga used to add to samples 8 Bit Octamed style 🙂

😀 😀 😀 😀 😀 😀 😀

EDIT* Hmmmmmmm and the joke is on me you said samplerate, I think it was 20.5 khz tops but i doubt i was using that, I was loading a lot of samples into 1 meg so i'm guessing probably a lot lower.

It has a certain sound that was used on a lot of classic anthems years ago by Force Mass Motion among others.

Shifrin · 10 Jun 2005

From memory, I think the sound chip (Paula) supported 4 channels of 8-bit samples at a rate of 20Hz up to 29kHz. (And a 6-bit volume register 😉 )

Shif.

Nu Audio Science · 10 Jun 2005

Cheers Shif now how do i replicate it 😀 😀 😀 😉

Shifrin · 10 Jun 2005

Nu Audio Science wrote: Cheers Shif now how do i replicate it 😀 😀 😀 😉

Export you audio and load it into a wave editor then resample it to 22050Hz 8-bit 😉

Oh, you wanted a way to do it in SM?! Erm... doesn't Nildon have a resampler/bitcrusher in it? You could play with that - then again you may have already done so!

Shif.

Nu Audio Science · 10 Jun 2005

Oi you ignoring me 😀

I said in the first post that Nildon wont load anymore 🙂

Nu Audio Science · 10 Jun 2005

Shifrin wrote: Export you audio and load it into a wave editor then resample it to 22050Hz 8-bit 😉

Shif.

😀 😀

Shifrin · 10 Jun 2005

My bad, didn't realise Nildon no longer loaded!

Shif.

unkargherth · 13 Jun 2005

mmmm 8 bits.....try this

polyin i; polyout o; float t1;

t1=0.5*(i+1); // make it polar t1=t1*256; t1=rndint(t1); // quantize it to one of 256 posible values t1=t1/256; o=2*t1-1; // make it bipolar again;

jorge · 13 Jun 2005

Shifrin wrote: My bad, didn't realise Nildon no longer loaded!

Which version of Nildon would that be? I didnt realise it didnt load anymore.

Nu Audio Science · 13 Jun 2005

.41 the one you modified and reposted 🙂

It hasn't worked since the arrays, I wouldn't worry about it too much at the mo though cause Steph is away 🙂

(I haven't even had chance to try Unks method yet - My Centrinos GPU blew up so i'm awaiting the return of that from Acer 🙁 )