Bit Depth, Sample Rate and Softsynths

[djsinistral]

What is the best sample rate and bit depth to export softsynths?

I know recordings at 44/16bit gain nothing by rendering to 96/24bit, but softsynths (I guess) will output up to 192/32bit, correct?

I specifically want to send stems (individual track bounces) for remix purposes. I usually use 96/24bit. But would it be better to go 32bit?.... using 24bit does appear to add a step into the export sequence where the rendered files are converted to 24bit... From 32bit? Then rendering to 32bit should ensure no downsampling is happening... Correct?

Thanks in advance.
.

[jlgrimes]

24 bit 44.1k for CD



IF synth have oversampling option be sure to try it.

[djsinistral]

24bit for CD?? CDs are 16bit for a start...

Thanks anyway.

Anyone else? (Who actually understands my question)

[TomViolenz]

If your project runs at 32 bit, then yes the soft synths will record in 32 bit.
If you want to send these tracks out as stems for remixing and the remixers will do their own downsampling to 16Bit CD quality, then yes the highest quality would be reached by doing the bitreduction only once.

These will be quite big files though.

[tedlogan]

I don't think you'd gain anything noticeable by a human being between 24bit and 32 bit. It's essentially just the volume's dynamic range is it not? Only at 8 bit can you start hearing some sort of noise floor, at 16 bit you've got a huge dynamic range, at 24 bit you've insane room for error. 32 bit? I'm not the most knowledgeable guy regarding this stuff, so someone please inform me what I'm missing.

[TomViolenz]

I don't think you'd gain anything noticeable by a human being between 24bit and 32 bit. It's essentially just the volume's dynamic range is it not? Only at 8 bit can you start hearing some sort of noise floor, at 16 bit you've got a huge dynamic range, at 24 bit you've insane room for error. 32 bit? I'm not the most knowledgeable guy regarding this stuff, so someone please inform me what I'm missing.

Yes exactly. So if he wants to preserve the maximum of dynamic range for his stems, then a 32 bit file recorded in a 32 bit environment will do that best.

If the difference really matters, is another thing.
But that's not what he asked

[tedlogan]

No he didn't ask that indeed, but I'm just trying to ascertain - why bother working in 32bit in the first place?

[TomViolenz]

No he didn't ask that indeed, but I'm just trying to ascertain - why bother working in 32bit in the first place?

Eh, more headroom?!

[tedlogan]

Isn't that like having a flea jumping competition inside someone's lounge, and then deciding "no, this won't do at all. We need to hold this competition inside a cathedral!"

[eyeknow]

I've heard every possible angle on this and though I personally find it a worth while conversation (as opposed to most) I no longer worry about it.

I run at 48k because I can hear a difference in ampsims and it gives me a bit better latency. I find that concentrating on what you have, in that context works better than obsessing over something that there seems to be no definitive answer for.

[deva]

I generally render softsynths at 96khz

With some it makes a significant difference and others a small difference.

[jlgrimes]

See next post.

[jlgrimes]

24bit for CD?? CDs are 16bit

24 bit refers to the format for mixing mixed down individual audio stems (what your subject seemed to imply).


The remixer or mastering engineer will take care of converting the final format to 16 bit.


32 bit would be great but possibly overkill for individual stems (unless you are bouncing multiple plugins effects in series and/or mixing layered synths to one file).

If you are posting this to the internet you can save some space by sticking with 24 bit. Bit depth is basically your dynamic range and 24 bit is huge. Only reason people (mainly DAW internal processing) use 32 and 64 bit is when mixing multiple tracks together and using lots of plugins errors can pile up and start to affect your audio audibly, but for individual stems (Raw synth tracks) this shouldn't be a problem.


44.1khz is a very common mixing format, so I would try to make sure the synths sound good at that rate. Synths with oversampling usually help here.

88.2khz would be another option if you are having aliasing issues using 44.1khz, but in the end it comes down to what format the remixer will use to mix.


I never had a mixing engineer complain that I sent them 24 bit audio files. That is pretty standard for mixing.

[jlgrimes]

I know recordings at 44/16bit gain nothing by rendering to 96/24bit, but softsynths (I guess) will output up to 192/32bit, correct?
.

Sample rate is an area you need to use your ears.

Synths can output up to 192khz (at least most of them), but many of them were tweaked to sound best at certain sample rates or oversampling settings. Some synths even specify this in their manuals like Monark.

I've had certain synths that sounded better at 44.1 kHz. It all depends on who designed them. Technically stuff should sound better at higher sample rates and they usually do (or make no difference) but not always.

[Tarekith]

Personally, I would never use 192kHz for anything. Period. In some cases it can be less accurate than a lower bit depth, and it's plain overkill anyway.

Generally I still render at 24bit for everything these days. 32bit is a nice safety, but I'm pretty sure my stuff isn't clipping so I don't need it. Most of the time I render softsynths at 44.1kHz and don't worry about it. Very rarely I'll render them at 96Khz just to see which I prefer, as was mentioned, with a few synths it can make a slight difference. Not better or worse I find, but it's interesting to compare none the less. Even if I choose the 96k version, I downsample it to 44.1kHz right away anyway.

Not sure exactly WHY they sometimes sound different at various sample rate, but there you go. I figure it's an aliasing thing most likely, so using 96k is pushing the aliasing up beyond Nyquist. That's one reason I don't mind downsampling right away to 44.1kHz.