The HD-Audio Challenge II

[ogs]

I recently had a very brief exchange with Morten Lindberg of 2l about 44.1 in 16 vs 24 bit (in Norwegian. The below is a short summary in English).
Lindberg said that for bandwidth alone 44.1 is OK, but that there is more to it than just bandwidth. If you look at the time domain, one needs higher sample rates than 44.1 to reproduce an impulse response without pre and post ringing. With high quality dither, 16bit may be sufficient, but he also said that 24bit is widely accepted as standard for studio use. As distribution is so easy via downloads, why not use 24bit for the end product too. Lindberg was more concerned with the fact that workstations often do floating point processing and conversion from float to fixed point introduces 'nasty' errors (without describing how these errors manifest themselves).