Has anyone ever used implemented dithering, used stuff like white noise, or different bit rates for anything?
My favorite thing to do with bit and sample rate crushing is in making percussion samples. Take tons of random field recorded hitting, kicking, drum sticking on objects and crush it. Stretch them thereafter for cymbals and toms. Add some filter to adjust harshness or make them more hatty or cymbally … though I usually don’t do that until they are in the sampler and often have an LFO on the filter freq and/or velocity controlling freq cutoff.
Use white noise in a lot of percussion too, often just a little supplement for a snare sound made from other things, processes.
Only use dithering settings in mastering.
Correct. Dithering is a very low-level noise that is used for mastering. Every DAW has it as an option in export settings, also most of limiters too - Ozone’s Maximizer, FabFilters Pro-L… It corrects quantization errors when you downsample while rendering audio. It’s pretty much inaudible and for example, Ozone also has auto-blanking mode when there’s no dither being applied if the signal is silent for at least 0.7sec. Dithering comes with different frequency shapes that’s why there are more options to choose from.
Now with using higher level noise… You can use pink noise a reference for mixing and it often suggest a neutral sound for your mix. It’s a cool way to find out which layer is louder than the rest. It really can help with balancing if you’re not too sure but keep it in mind it’s not always the greatest solution.
Noise in sound design, lots of options. Layering, creating transients for drums… Completely different topic.
I use SpecOps’ MP3ify, if that counts. Usually when I’m doing my Alex Jones impressions
Dirtying up any sound with artifacts like that gives me the jollies for some reason. It’s like your normally-recorded stuff can instantly sound like you sampled it from someone else. “Where did that sample come from?” they’ll ask. A slight grin will spread across your face.
I hear somewhere people talking about mixing with White Noise but did not attach any importance to this.
I hear that is better to use 96kHz than 44100 when mixing and export to 441000 Hz.
I hear that you can apply dithering in mastering stage.
Dithering is always used for mastering. General rule is always add dithering when your product is ready to go, it exists for a reason. 96kHz stuff is mostly used by recording companies that does foley/field recording. The reason is that most of customers always end up downsampling their product, every time you apply something, like an audio effect, it’s getting downsampled and common user is often in 44.1k/16bit settings because most of the time it’s default DAW settings and there’s nothing wrong with that. 96kHz is that RAW realm, it exists just in case, otherwise 44.1k is perfectly fine for music, that’s why it’s always the default option and trust me, you won’t hear any difference, if you do, it’s just placebo in action.
Noise is an artistic thing for me, I don’t use it for mixing unless I get super lost, then I’ll use pink noise to get me back to a decent balance.
Now, dithering and bit rates…
So for dithering, I use it on my masters, 24 or 16-bit. I keep all working versions of tracks, any stems I render, samples, etc. in 32-bit and don’t worry about dither there (though technically that is possible now, but its a very recent thing and I question the benefits at that level, which is like 800 db down or something insane). For the 24 and 16-bit masters, those get dithers. I use Airwindows Not Just Another Dither because it shapes the noise very well in the frequency domain, but also the time domain. It basically gives you just enough noise to cover up the quantization artifacts at any given time, which helps it get out of the way of music even more than most noise-shaped dithers. And it’s free.
Call me crazy, but I don’t worry about the sampling frequency side of bit rate much (perfectly happy at 44.1khz), but I do care about the bit-depth side of things. I prefer 24-bit music where I can get it (and it’s what I render as my default, lossless master), just because it does push the noise floor down a bit.