Thread with 4 posts
jump to expanded post
lmao 32-bit integer linear pcm exists. who would ever actually use this. 32-bit float forever
kinda curious what 16-bit float audio sounds like, i should implement an audio processing system based on 16-bit floats just for the hell of it :3
@hikari if 16 bits floating point is still way too precise to have any audible effects, there's also the whole topic of https://en.wikipedia.org/wiki/Minifloat with formats that follow IEEE 754 as well as formats that don't
@hikari 16 bit is just about the point where the noise floor is barely audible (no noise shaping) or inaudible (noise shaping quantization)
i'm not sure what introducing the weirdness of float does to this but it sounds like a fun exercise to find out