I use a MixPre 6ii. I record 32bit float/48kHz. I bring the polywav file into Reaper. I explode the polywav to multi-track... and away we go on processing.
My question is:
Do I need to apply dither to my renders? I am rendering to 24bit/48kHz wav to later make flac in TLH. Do I just need to apply dither if I make an additional 16bit/44.1kHz copy?
i always dither, rather than just let the software save or convert while saving ....
I don't think I ever noticed that after I explode the polywav in Reaper, the tracks are still 32 bit. How I did not know this was native behavior is an oversight, even after 6 months of taping shows with it.
My thinking was dithering was always for rendering kHz rate and not about bit rate so I have only dithered when moving from 48kHz to 44.1kHz when making a "lite" copy for some of my friends' bands.
It is my understanding that, when reducing bit depth, best practice is to dither to reduce distortion due to quantization error. I don't know whether it is strictly necessary, but it is easy to do, so why not?
Quote from: Twenty8 on May 23, 2023, 08:59:54 AM
I don't think I ever noticed that after I explode the polywav in Reaper, the tracks are still 32 bit. How I did not know this was native behavior is an oversight, even after 6 months of taping shows with it.
My thinking was dithering was always for rendering kHz rate and not about bit rate so I have only dithered when moving from 48kHz to 44.1kHz when making a "lite" copy for some of my friends' bands.
The terms typically used for these two different processes are -
dither - changing bit rate
sample rate conversion - changing sample rate
Here's a good non technical article from Izotope on what these processes are and when they should be used - https://www.izotope.com/en/learn/what-is-dithering-in-audio.html (https://www.izotope.com/en/learn/what-is-dithering-in-audio.html)
I feel like the implication of the article suggests two renders: one to dither to 44.1kHz and a second for sample rate conversion to 16bit.
Is that the correct way of reading the article?
Quote from: Twenty8 on May 24, 2023, 09:32:09 AM
I feel like the implication of the article suggests two renders: one to dither to 44.1kHz and a second for sample rate conversion to 16bit.
Is that the correct way of reading the article?
Dither doesn't apply to sample rate conversion.
You can apply dither (to reduce/change bitrate) and resample (to convert from 48kHz to 44.1kHz) as part of the same render so only one process. I haven't used Reaper in a while but in Wavelab (and other DAWs) it's one process with selections for final format made at time of render.
Is it bad that I have NEVER dithered and record at 32 bit, compress to FLAC 8/24 and upload? I have never had an issue, this is in Audacity BTW....
Dithering has zero effect on sample rate conversion. Dithering when converting sample rates won't really hurt your tape, but it will add time to the bounce/render when you apply it.
In theory, you want to dither whenever you change bit depth. It helps to mask strange distortions during fade ins/outs, and it technically masks some (usually imperceptible) artifacts when volume is very low. In practice for taping, it doesn't make a huge difference whether you dither or not. I always dither when going from 24 to 16 bits, if you're going from 32 float to 24 I'm skeptical anyone would hear the difference between dithered and not dithered.
A good exercise is to try bouncing two files from 32 to 24 bit - one dithered, one not dithered. Especially on fade ins/outs. See if you hear a difference. When I was younger and my hearing tested up to 23k, I could hear it but decided even back then it was a subtle thing that I'd do for polish but wouldn't stress if I forgot to add in; these days with my hearing only going up to 17k I'm not convinced I could hear it.
I record 32FP/48kHz and final tracks are 24/48 FLAC, same as BlueSky71. My typical post workflow is iZotope RX (global level adjustments, spectral denoise, other corrections) then Reaper for everything else. I keep everything in the 32FP until final track rendering, and never add dither, unless making 16-bit tracks for CDs. And I haven't done that for ages.