16 bits or 24 bits ?

[Sharmaji]

24 bit makes an audible difference, no doubt.

[Littlefoot]

USE 24/44.1 AND GET ON WITH MAKING TUNES

[macc]

USE 24/44.1 AND GET ON WITH MAKING TUNES

One question though: will you sound card strap a different anti aliasing filter across the input when you switch to another sampling frequency?

Yes, it's arguably the main reason for using higher sampling rates. It can use a more relaxed filter, which is easier to implement, as the aliasing is shifted up the spectrum out of the audible band.


<----- I like this dude.

[hurlingdervish]

does 24 bit (or is it higher sampling rates im thinking of) give you less degradation or even better quality when processing and resampling?

ie, pitching up, then back down, etc etc

[drdeft]

THanks for the answer !

Nobody working with 24/96 ?

[macc]

does 24 bit (or is it higher sampling rates im thinking of) give you less degradation or even better quality when processing and resampling?

ie, pitching up, then back down, etc etc

Read the thread

[hurlingdervish]

does 24 bit (or is it higher sampling rates im thinking of) give you less degradation or even better quality when processing and resampling?

ie, pitching up, then back down, etc etc

Read the thread

i did thats why im asking!

im not talking about just the way it sounds after recording, i mean the way it sounds after recording and resampling like 10 times

or purposefully giving it aliasing by pitching up and back down

[macc]

I'll just copy what I wrote earlier In combination with the anti aliasing comments above you can fill in the blanks

In terms of bit depth, the computer will populate the new, empty bits (y axis, essentially) with zeroes when changing a 16-bit sample to 24-bit. This does not ‘give a smoother sound’. It makes no difference whatsoever to the sound at that point. What it does do is enable subsequent calculations/processing to be done with a lower noise floor, ie more accurately. It doesn’t give you anything back, that’s already lost. It just helps to stop things getting more f#cked up when you process further.

For example, if you get a great 24-bit drum loop, resample it at 8-bit, and then resample that at 24-bit, you don’t get the initial drum loop back. You have an 8-bit sounding version, but at 24-bit resolution. This principle is one reason why most plugins work at a higher internal resolution than the native resolution, it reduces the noise floor for that stage of processing.

When it comes to sample rate, it’s not dissimilar when over/upsampling – here the gaps between samples are filled with zeroes, and it doesn’t do anything inherently to the sound. Loads of higher frequencies that were lost at 44.1kHz don’t come flooding back. But computers don’t do this by default, they work at a native/host sample rate and any upsampling etc is handled internally by plugins (when a plugin upsamples). The purpose is much the same as with the bit depth case, better accuracy, but in this case it matters more at the high end of the spectrum, be it for the purpose of HF decramping in eqs, or for better compression response/handling etc. Point being it still comes down to the native sample rate before being passed back to the host, just with potentially less inaccurate results.