Question about bit depth

I have several projects that were originally recorded at 44.1KHz, 16-bit. If I import those files into a 44.1KHz/24-bit project will it screw up the project? Should I just finish those projects at their original bit depths?

I’ve never experienced a problem with my DAW converting the tracks. I do know there can be issues if you DON’T convert them :wink:

Are you using Reaper, Al? I know Reaper and Sonar convert sample rates automatically, but I haven’t had any experience with different bit depths.

I am using Cubase. I knew I’d forget an important detail. lol

nothing bad will happen. Everything is converted into 32 or 64 bit floating when you are mixing anyway.

5 Likes

Interesting. Thanks, Boz!

That is what i would have said had i actually had the knowledge. :smile_cat:

2 Likes

Hell no nothing will hapopen.

If I render down a midi track into a wav, I often forget to set it to 24 biut so it defaults to 16.
Plays perfectly well and I’m buggered if I can tell the difference.

Thanks, Shack. The way it was explained to me, the bit depth gives you more “mathematical room” when mixing. In other words, you’d hear certain artifacts working in 16-bit that would be inaudible in 24-bit. This stuff is giving me a headache. :beerbang:

1 Like

Yeah, 24 bit is mostly about giving you leeway. 16 bit has 96dB dynamic range, so the quietest signal it can meaningfully reproduce is down there at -96dB. Next time you’re mixing, turn your master fader down to -96dB to see how quiet that actually is.

Of course, the signal at -96dB will only have 1-bit resolution, so it’ll likely sound pretty rubbish - in crude terms, it’ll turn a sin wave into a square wave. Its only saving grace is that this distortion will be so quiet, it’ll be inaudible on most playback systems, which is why 16-bit is fine for CD.

When you’re mixing though, things are a bit different. For starters, A/D converters have a range they’re designed to operate in - typically somewhere around -18dBfs = 0dBu. So for a typical signal you record, you’ll not even be using the top 1 or two bits, leaving them as headroom for unexpected peaks. Each extra bit = 6dB of extra dynamic range, so if your signal is peaking at -12dBfs in the computer (a good level) and you’re recording at 16 bit, you’re only effectively using 14 bits and your dynamic range is now 84dB, which means relatively speaking the lowest significant bit moves up in volume closer to the level of your signal.

Still that wouldn’t be too much of a big deal, if you weren’t going to do anything with the track you record other than maybe normalise it; -84dB is still reasonably quiet. But when you mix, you mangle the track in all kinds of ways;

Say it’s a rock vocal and you heavily compress it by 12dB gain reduction. You use makeup gain to recover the level, and now your distorted signal around the least significant bit is boosted to -72dB, which you might start to notice as grainy ‘digital’ harshness. A few words are still getting lost in the mix so you automate a few fader boosts and that brings the noise up too. You EQ a high end boost and that least significant bit square wave hash gets a free boost as well. You decide digital is hell and vow to only ever record to cassette tapes going forward. :grin:

By contrast, 24 bit recording gives you 144dB dynamic range. You could record with peaks at -48dBfs and still get 16-bit, CD resolution. There’s enough dynamic range that most analog gear can’t even match it - typically anything below -110dB or so is just thermal noise, electrons bouncing about in the circuit.

As @bozmillar said, DAWs work at 32- or 64-bit floating point anyway, no matter the resolution of the incoming audio streams. So if you have a 16 bit file and add reverb, the reverb trail will be at the DAW’s own working resolution. If you process the 16 bit file, it will become a 32- or 64-bit stream, so you won’t make things worse or lose any more meaningful information. But the DAW’s mix engine won’t be able to restore any information that was lost during the 16-bit recording, so any distortions that were recorded are there to stay and can be accentuated as per my example above.

Here ends my first (probably unwanted) essay. :joy:

5 Likes

No way, this is great stuff! Thanks for the very clear explanation. I think even I finally get it now. :relieved:

1 Like

Just the opposite. I am THRILLED that you wrote this essay. Thank you so much.

1 Like

In Cubase, when you import an audio file (or just drop it in your timeline), it will ask you if you want to convert it to the current sample rate/bit rate (you can set this in your project preferences).
Just say yes, and you’re done.

1 Like

I truly wish I had a more technical brain. I get there eventually but it is hard fought. Good info in this thread:)

I liked the part about “the rest is just the electrons bouncing about in the circuit”.

1 Like

Good explanation.