at a certain point i lost the interest in making music - somewhere in my early 30's.

[lritter@mastodon.gamedev.place]

@aeva suggest a way

what kind of math do you want to do?

[aeva@mastodon.gamedev.place]

@lritter audio math

[lritter@mastodon.gamedev.place]

@aeva welcome in my present cell. the food is terrible and the view is even worse.

no i joke, i'm just drawing a blank at the moment. i'm now trying to relax, do something else and hope for an epiphany to catch me.

[lritter@mastodon.gamedev.place]

@aeva it would probably be easier to enumerate the shit ways to do audio math in datalog and by that figure it out.

but first, i need crepes in my stomach.

[lritter@mastodon.gamedev.place]

@aeva name one specific kind of audio math

[tomf@mastodon.gamedev.place]

@lritter @aeva x = (x + 1e-36f) - (1e-36f)

:-)

[lritter@mastodon.gamedev.place]

@TomF @aeva that's only interesting for boundary conditions. doesn't sound very nice.

but fine:

t(i1,x1) :- t(i,x), fadd(xx, x, 1e-36f), fsub(x1,xx,1e-36f), iadd(i1, i, 1).

[tomf@mastodon.gamedev.place]

@lritter @aeva You shouldn't hear anything. It's to remove denormals from the audio stream, which can produce 100x speedups on some CPUs.

[lritter@mastodon.gamedev.place]

@TomF @aeva silence doesn't sound very nice.

also y u no fixedpoint?

[tomf@mastodon.gamedev.place]

@lritter @aeva Audio is a PERFECT use for floating point, because the value 0 absolutely means something special, and having more precision near it is very useful.

[lritter@mastodon.gamedev.place]

@TomF @aeva why is it useful?

[farbs@mastodon.social]

@lritter @TomF @aeva The nonlinear relationship between perceived loudness and signal amplitude means precision is more important for quieter sounds, and audio signals are usually centred around zero to reduce clipping.

[lritter@mastodon.gamedev.place]

@Farbs @TomF @aeva what does that matter after a 16bit mixdown?

[farbs@mastodon.social]

@lritter @TomF @aeva
Oh yeah. Whoops. Looks like I was wrong there, sorry.

But it does matter before and during mixdown (DSP and dithering).

[lritter@mastodon.gamedev.place]

@Farbs @TomF @aeva why?

[tomf@mastodon.gamedev.place]

@lritter @Farbs @aeva So that you can have intermediate volumes during the mix that won't introduce clipping or quantisation when having a second volume applied afterwards.

But you know this - you have made plenty of music.

[jkaniarz@mastodon.gamedev.place]

@TomF @lritter @Farbs @aeva I’m not so sure float is actually improving anything. 14-bits is all you need for the final mix. 24-bits lets you record a whisper and a jet engine without having to adjust the gain between takes. 32-bit lets you have all that plus room to cover accumulating quantization error in you DSP chain. Fixed point is fine. I think it’s more about hardware performance or developer convenience than sound quality.

[nickappleton@mastodon.gamedev.place]

@jkaniarz @Farbs @lritter @aeva @TomF I think your comments around performance and developer convenience are spot on, but writing fixed point code is awful for anything of significant complexity. Cascading IIR filters with low frequency poles leads to complexities in analysing headroom requirements. Mixing large numbers of channels requires level analysis. And if processing is occurring in some filterbank domain, then you have to consider fixed point FFTs.

[nickappleton@mastodon.gamedev.place]

@jkaniarz @lritter @aeva @TomF @Farbs while writing fixed point code is “fun” from a puzzle-solving perspective… I am really looking forward to not needing to do it anymore.

[tomf@mastodon.gamedev.place]

@nickappleton @jkaniarz @lritter @aeva @Farbs Right. I'm normally "Mr Fixed Point" but in this case zero has a very good meaning, and having highly dynamic range around it is very valid and useful and solves a lot of annoying problems.