Topic: 16 vs. 24 bit recording which do you prefer

[Gutbucket]

I think the switch to recording 24bits may have been more significant in marking a change in taper behavior by encouraging more conservative level setting than marking a significantly meaningful change in gear capability.

If I get an extra bit or two of range out of a recorder with an ADC stage that can only manage 17 or 18 bits at best when writing 24bit files, I'm fine with that.  I don't worry much about having levels too low since I've I don't think I've ever had my gear's noisefloor end up higher than that of the recording environment. 

There is a local chamber music outfit who have performances in a small general purpose meeting hall here.  The room is rather mediocre to begin with, adding nothing positive accoustically, yet the overwheliming negative by far is the rattling AC vents in the drop-ceiling.  French doors to outside front and back patios are kept open before the program and at intermission, so the AC is always raging and rattling non-stop through the entire performance.

[H₂O]

Record and seed in highest res practical - I try and seed the recording as close to orig as possible - no edits unless a channel drops, etc - same bit depth and sampling rate

Only exception is DSD recordings where I seed at 24 88.2

I am a purest though

On LMA I have LMA create mp3's for causual listening

[cashandkerouac]

If and when I tape, I do so 24/48 (why not???)...

When I Seed, I do so at 16/44 (convenience for end-user)...

When I listen, I mostly do so via MP3 (crappy headphones at work and in the car) so it doesn't matter... 

Terry

similar for me, except for the fact that i record at 24/96.  i don't hear an usible difference between 24/96 and 24/48, but storage space is cheap so i figure it can't hurt to record at a higher resolution. 

[Stagger]

I record in 24 bit because my recording and playback gear can do it and it makes me feel better about myself as an audiophile. Hard Drive space is cheap.

However, given the specs and the actual dynamic range at rock concerts I readily acknowledge that there isn't any difference in sound.

I think those who claim to hear a difference should be backing up their claims with ABX results.

I think it depends on what you are recording and how advanced you want to get with your post processing. The wealth of plugins that are out there that, when properly used can make use of headroom beyond the originally recorded dynamic range is vast and ever expanding... All the software children of the Aphex Aural Exciter. You can restore compressed harmonics and even compressed dynamic range in general, if that is your wish. Even with large PA recording, you may have sections of dynamic range that are much greater than your average. If you are recording acoustic (or at least not fully reinforced) music, there can be a pretty clear difference.

Backing this up is harder than simply comparing a resampled 24 bit recording though. What we are really talking about is what happens at the point of capture. You would have to basically run a rig split at the AD point with everything the same save for the bit depth which, unfortunately, I don't have the ability to do. 

[DSatz]

Jon, due to a setup error on my part, I once made a 16-bit live classical recording without effective dither, and the results weren't what I expected from general theory. The noise floor of the venue included the building's ventilation system and Manhattan's omnipresent traffic rumble, so I thought that the recording would be "self dithering." The lowest levels I saw on the meters were ~54 dB below full scale, and since the theoretical requirement is only that the LSB be fully randomized, I thought that there would be no problem.

But when I got home I found that the recording had quite audible granular noise (quantization distortion) on the fadeouts of sustained piano tones and even on some quietly sung vocal passages. It turned out that those -54 dB noise floor levels were due in large part to uncorrected DC offset in the A/D converter. Once that was subtracted out, the lowest levels were more like -72 dBFS, in part because the preamp gain was set lower than would have been optimal.

The preamp + A/D was a Grace Lunatec V3, and I'd forgotten to engage the 16-bit dithering function. From this experience I learned always to double-check that setting before recording. Also, once I realized what had caused the falsely reassuring level readings, I sent the V3 back to Grace, who had worked out a method to get rid of DC offset since my preamp had been made. In general I consider small amounts of DC offset to be harmless, but not if they throw off your estimate of the noise floor. (I'd set my overall record levels based on what I thought was the noise floor when I was setting up, and since that reading was artificially raised by the DC offset, my levels were lower than they would otherwise have been, thus exacerbating the problems caused by the lack of effective dither.)

Obviously these were practical mishaps and mistakes that don't disprove the theory, but they do show how mistakes and mishaps can get the upper hand in the real world.

--best regards