[louthy]

> For audio production and mixing 24/96 makes a lot of sense though, because of dynamic range.

Obviously only the bit-rate matters for dynamic range, and a 16 bit signal has 96 dB of dynamic range. That is more than enough for even the most dynamic of audio signals.

A larger bit rate is useful for lots of digital summing as I mention here [1] and I assume what you allude to, but for most home applications nobody needs 16bit+ for anything other than improving their noise floor (which is still borderline inaudible at 16bit).

[1] https://news.ycombinator.com/item?id=16995020

[squeaky-clean]

> A larger bit rate is useful for lots of digital summing as I mention here [1] and I assume what you allude to, but for most home applications nobody

Yeah that's why they said audio production, and not home use. 96db of dynamic range sounds like a lot, but not when you're stacking 45 tracks. Also compression and saturation later in the chain will further bring up that noise floor.

And you don't get that full dynamic range because you want to prevent clipping in a recording, so you give yourself something like 12db of headroom away from the maximum, and now you've lost a good chunk of your dynamic range. 24 bit recording lets you keep plenty of maximum headroom while also staying far from your noise floor.

[louthy]

> Yeah that's why they said audio production, and not home use.

Yes, but the comment was specifically about dynamic range in audio production, not about higher bit rate's prime benefit of reducing artefacts in digital summing. The noise floor at 16 bit is -96db, if you're down there, you're doing it wrong.

> 96db of dynamic range sounds like a lot, but not when you're stacking 45 tracks. Also compression and saturation later in the chain will further bring up that noise floor.

It's more than enough if you're doing proper gain staging - the number of tracks is irrelevant.

Any mastering compression of more than 3db gain reduction is mostly excessive, but let's say 6db. So a final master will bring up a noise floor 18db at most (if gain staging was done properly and as you say you left plenty of headroom for the mastering engineer - although clearly you wouldn't be able to use all the headroom). That's still a noise floor of less than -78db, but more realistically around -85db.

Saturation usually only add harmonics not gain (obviously depends on the kit you're using).

> And you don't get that full dynamic range because you want to prevent clipping in a recording, so you give yourself something like 12db of headroom away from the maximum, and now you've lost a good chunk of your dynamic range. 24 bit recording lets you keep plenty of maximum headroom while also staying far from your noise floor.

Myth. Watch this [1] and you'll realise that there's plenty of range with 16 bit. People were producing with 16 bit for years perfectly fine. Before that high end studio tape machines (like the classic Studers or Telefunkens) were equivalent to about 14-16 bits (based on their noise floor) and we had decades of music produced and recorded on that format.

I am not saying that 24 bit isn't better than 16 bit. Of course it is, there are tangible benefits in summing and it gives you more headroom to work with. But, I'm saying it's a myth that there's not enough dynamic range or that you're somehow just bumping along above the noise floor. If you are then you're doing it wrong.

[1] https://www.youtube.com/watch?v=cIQ9IXSUzuM

[lozf]

Another much appreciated advantage of 24 bit is that it allows so much extra headroom:- when recording, levels can be set so the peaks hit maybe -18 -- -12dB(FS) which helps eliminate any chance of clipping in unexpected circumstances.

[squeaky-clean]

> Yes, but the comment was specifically about dynamic range in audio production, not about higher bit rate's prime benefit of reducing artefacts in digital summing.

But audio production often involves lots and lots of digital summing. I don't see how that's not related.

> Any mastering compression of more than 3db gain reduction is mostly excessive, but let's say 6db. So a final master will bring up a noise floor 18db at most (if gain staging was done properly and as you say you left plenty of headroom for the mastering engineer - although clearly you wouldn't be able to use all the headroom). That's still a noise floor of less than -78db, but more realistically around -85db.

That's only counting mastering compression.

What about NY style parallel buss compression on drums where you want to hit -20db GR as an effect, which will inevitably add some saturation too. 6-10db of compression on an individual drum track is common too if you're doing metal or electronic music. So you've got 10 drum tracks, each with several effects and a parallel track with distortion/compression.

> Saturation usually only add harmonics not gain (obviously depends on the kit you're using)

If it's not adding amplitude to your signal it's almost certainly bringing up the noise floor. Saturation generally works by applying gain until the device/plugin distorts. Just because amplitude isn't going up doesn't mean gain isn't increasing.

> Myth. Watch this [1] and you'll realise that there's plenty of range with 16 bit. People were producing with 16 bit for years perfectly fine. Before that high end studio tape machines (like the classic Studers or Telefunkens) were equivalent to about 14-16 bits (based on their noise floor) and we had decades of music produced and recorded on that format.

This is true for when you're operating on friendly signals like a sine wave signal generator with a fixed amplitude. Not so nice when you're tracking with 20db of headroom and the drummer some how manages a flam 24db f*cking louder than everything else which clips the overheads (this actually happened using 24bit, but now with crazy dynamic drummers I give myself TONS of headroom because there's no reason not to). At least when tape "clips" it doesn't ruin the take.

I do agree with you that 16bit can be good for recording (and for home audio I agree you'll never be able to tell a difference) but I don't believe there are literally zero differences between 24bit, and hard disk space is so cheap that there's no reason not to use the extra bits.