What a surprise. oggenc encodes Mussorgsky significantly faster (in input bits per second) than it does Bartók
Current Mood: amused Current Music: Queen -- Somebody To Love Entry Tags:geeky, music, one-liner
It's a surprise to me.I'd have assumed that the encoding was a FFT derivative, quantisation and then some sort of variable-length coding - all of which are substantially constant-time operations for each input bit.
I guess it must have trouble managing the total output bitrate and needs several passes but I'd have thought that could be managed in two complete passes - keep a priority queue, possibly one for each block of data, with the priority determined by the effect that allocating an extra bit would have on the RMS(?) error...