So I record my projects at 32 bit float, then I dither to 16 bit on the stereo output for mixdown but I’ve only just noticed that when I’m exporting a mix the audio engine output is still on 32 bit float, so does this bugger up my 16 bit dithering
ARGH! I get headache every time someone says: “dither to”. You don’t “dither to”, you “reduce bit-depth to” and you can do this with or without dither.
Funny: I get a similar headache when people refer to word length as “bit depth”
On a more helpful note, there’s a very useful explanation of how and why dither may works in this SOS feature, complete with audio examples: Digital Problems, Practical Solutions
yes I was aware of that,I was just concerned that upon export my audio engine was set to 32 bit float and if this effected anything… thank you for popping in though.
You’re free to simply convert those mixdowns into 16-bit. There shouldn’t be any information past 16-bits—as long as 16-bit dither was the very final process.
I can see your point. Word length is technically right term. Anyway I refuse to use it in context of number of bits in PCM sample, because it may be confused with word length of computer’s CPU. Maybe “sample length” could be reasonable compromise.
I was only gently teasing… Everyone knows what you mean by ‘bit depth’. Best not start talking about sample length, though, as that has even more potential to confuse!