Wouldn't recording to tape at significantly below 0 db significantly lower the signal to noise ratio?
Tim
sure would... then low signals like reverb tails and spatial ques will get lost in the tape noise.
I'm puzzled about you saying PCM loses bits, as far as I was aware there is a tradeoff between sampling frequency and measuring absolute level between the 2 techniques, but if the filtering is doing its job properly there should be no loss of bit encoded information.
Frank
Sorry... I'm not an EE. I'll leave the techno jargon to Don and others who can explain it better.
I'll include a quote from the other forum and let the EE guys tell me if it passes muster.
from bdiament:
"I see a lot of posts about the limits of human hearing above xkHz, yet I believe the advantages of wide bandwidth are not about supersonics but reside well within the audible range.
Even with purely analog signals, KEF showed decades ago, how proper time response (i.e. rise time, settling time) is tied to wide bandwidth. According to their work, proper time response at frequency x requires a bandwidth of 5x.
My own experience has been that wide bandwidth makes for more convincing reproduction and that bandlimiting doesn't. I would have to hear exceptions to this before anything else would make sense to me. So far, I haven't heard a single exception.
That's the "horizontal". Now to the "vertical", the number of bits.
Unless one is going to compress musical dynamics, some parts of real music are not as loud as other parts. The quieter parts might be 20, sometimes 40 dB down from the loudest peaks.
With a 16-bit system, this means the quietest parts will utilize 13, sometimes only 10 bits of "resolution". Instrumental (and vocal) harmonics are down in level from the fundamental notes being played (or sung). These harmonics will therefore be represented by considerably fewer than 16 bits. Anyone ever listen to a fine cello recorded at 8-bits? As I've said elsewhere, it is well on its way to sounding more like a kazoo than a cello.
Spatial cues too, are considerably lower in level than the sounds from the instruments and voices making the music. With this in mind, I don't find it surprising that comparing the same recording at high res vs. 16/44, it feels like the lights in the room have been turned off and most of the air sucked out of the space.
Now lets look at those same parts represented with a 24-bit system. The parts that are 20 dB down from maximum will utilize about 21 bits. The even quieter parts at -40 will utilize about 18 bits. We still have more resolution than the loudest peaks on a 16-bit recording. In my experience, this is plainly audible on any decent system. The quietest parts on a dynamic recording (the most difficult for digital to capture faithfully) are reproduced with appreciably greater fidelity on a 24-bit system than the loudest (easiest) parts are on a 16-bit system. For me, 24-bit decisively wins the race"