Convert 44.1k to 48k ór 16bit to 24bit

The Lounge General Discussion
#1

Asking cause I have some files that I need to batch convert to 24bit 48k for Bitbox Micro. But also asking out of interest to learn from. The internet is full of half related information, but I couldn’t find info on this exact comparison.

I have a bunch of samples that have two sizes: a 48k+16bit folder and a 44.1k+24bit folder. [Added: and I need to convert one of the two size types to 48k 24bit]

Does it matter, does it make a difference between up-converting 44>48kh vs up-converting 16>24 bit-depth? Or same? (The digital audio info isn’t there to begin with in either case, so maybe both are the same as in they can’t bring back information that isn’t in the source file to begin with?)

#2

If you have iZotope RX, that has a good batch converter.

For something free, I have also used fre:ac in the past https://www.freac.org/

#3

Edit: Sorry misunderstood - I would convert the 48k 16bit if the dynamic range and snr ratio is not an issue in the samples.

Are you sure the BB micro won’t play the 16bit files?

#4

Wavelab !

#5

I have a fine converter. The questions is which of the two options in my post?

#6

The conversion engines in both cases work the same way i imagine, like increasing pixel density, it’s still artificial but depending on the conversion engine it’s more or less good…

1 Like
#7

Either works just fine, there’s not going to be a clear benefit. In each case you’re taking food and moving it to a larger container. It’s still the same food, the container is just bigger.

It’s when you have to put the food in a smaller container that things get interesting, then you have to choose whether you sacrifice some of the mashed potatoes or the green beans.

3 Likes
#8

I would take the 48k 16 bit so there’s nothing to be recalculated in the bitrate. I don’t know any reference that confirms this but this seems to be the safest way not to loose anything (like pitch) only adding bitdepth. Lowest risk to decrease quality if there would be any.

3 Likes
#9

Apologies, I hadn’t had my coffee when I replied earlier!

The Bitbox Micro looks to natively support 16, 24 and 32 bit @ 48khz, so using the natively supported 16 bit files means you wouldn’t need to do any conversion. Reading over the support forum and modwiggler, there seems to be reports of aliasing artifacts introduced when it tries to handle sample rates other than 48k.

1 Like
#10

Yeah the BB really seems to want 48k. With 16bit it also has to do live re-converting it to 24bit, so in light of keeping the latency/lag down as low as possible I want to work with 48k 24bit with all my samples. Thanks for thinking along though, appreciate it

#11

I’d probably go with upping the 16bit 48k to 24bit, as that should have no impact at all in the audio, going from 44.1 to 48 can cause interpolation artifacts in the higher frequencies as the conversion has to add/create sample rate data that doesn’t exist.
The source files are probably already normalized when they were downsampled to 16bit anyway, so the only trade-off on the 16bit files is they’d have a slightly higher noise floor than the 44.1 24bit ones.
Unless the source samples are all low volume/high dynamic range, 16 -> 24 is what I’d do personally :slight_smile:

1 Like
#12

That’s multiple votes for using 48k 16bit to convert to 24bit.

The source files are from a commercial (and well recorded) pack, and indeed are already normalised. But that’s a good criterion then indeed - wether they’re sensitive in the headroom department or not.

Plus I have some nice reading up to do about interpolation artifacts!

Thanks everyone, I’m going for the 48h16bit>24b approach

2 Likes