Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It only takes a minute to sign up.
Sign up to join this community
I mix several tracks recorded with a bit depth of 16 bits.
In Audacity, I set the 32-bit setting to floating point. I don't change the sample rate.
Is it worth it to edit with the 32-bit floating point precision if the source files are 16-bit? Or it is better not to upscale the bit depth while mixing? The final mixed file I always save as 16-bit.
The advantage of working at a much higher bit- or sample-rate internally than either your initial input or final output becomes more & more apparent as you start to rack up more channels, group busses & effects plugins.
Each of these at 16-bit will add noise/randomisation/dithering right down at the bottom few bits of the process… each & every time.
If you are running internally at 32-bit, this is dropped factors lower in the actual perceivable audio, and by the time you save your final mix is likely buried in just those bottom couple of bits in your resulting audio file, making it, in effect, "noise added just once" rather than every time.
Many DAWs these days operate at 64-bit float internally [without the user even being aware]. Audacity's default setting of 32-bit is as high as it will go. Use it.
The only reason not to use it would be if you have extremely limited disk space - though personally I'd be looking at bigger disks if that was the case.
If the file remains at 16bits, each step of your mixing could add a unit of amplitude and a noise (around -96dBFS) will accumuate.
Then it is better to work in higher resolution. At 24bits (fixed) or 32 bits (float), the noise which accumulates will be around -144dB per step, which will be negligeable. The resulting noise will be around -96dB (the final conversion), not higher.
