[music-dsp] why 3/4 in quantization?
decoy at iki.fi
Mon Jan 19 16:34:00 EST 2004
On 2004-01-19, Joshua Scholar uttered:
>But I think it comes down to having some sort of instantaneous nonlinear
>quantization (is that what they call it?).
Or compansion, though that's a bit misleading if applied with a single
sample delay. As you say, that's what things like A-law and mu-law use;
NICAM (Nearly Instantaneous Companded Audio Multiplex, in case anybody's
interested) uses longer blocks but the principle is the same.
Hmm... One explanation I can come up with is that 0.75 might just be the
closest exponent to ~0.6 (in logarithmic units) which we can speed up with
bit twiddling tricks. The latter constant appears when we go from phons to
sones, that is, from a logarithmic equal loudness representation to
perceptually uniform units. (It might also be they've adjusted the
exponent upwards since I last checked.)
Alternatively I have an overactive intuition.
Sampo Syreeni, aka decoy - mailto:decoy at iki.fi, tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
More information about the music-dsp