“I really dig the 16 bit Naturalize. It has made me wonder in surprise if I like the reduced version better than the 24 bit original.” Andrew Levine, Blumlein Records
DitherboxDemo is a universal binary AU plugin supplying a wide range of unbeatable dithers! You get the traditional flat/TPDF plus High Gloss, Vinyl, Spatialize and Naturalize, plus three special monitoring modes.
First, you get raw truncation, flat and TPDF. TPDF is always ‘correct’ and with the demo you can shoot it out against the other ones, use it as a reference to what dither is normally like. Also investigate truncation—you’ll find crazy things happen at very faint levels, you get a snarling raspy hypercompressed version of the audio until it sputters out entirely.
High Gloss comes from experiments trying to make a more regular noise source. It’s designed to combine the noise floor with that hypercompressed effect truncation has, but tailor that into a more usable sound that draws attention to the textures of sounds. I see this as being suitable for some electronic music, not so much acoustic or hi-fi.
Vinyl is a seriously unique dither. It’s more CPU hungry, which can’t be helped (still nothing too outrageous though). This dither literally tries to suppress the actual noise floor, particularly in the highs, resulting in a sort of sputtery crackly effect that’s actually quieter than TPDF is, as well as being a lot more like record surface noise. It’s phenomenal at 16 or 24 bit- very pleasing listening.
Spatialize was originally Contingent Dither—I’ve named it Spatialize as that’s what it’s good at compared to the others. It throws a really detailed soundstage and is one of the top hi-fi picks. It also conveys air and high frequency information very well.
Naturalize is something a little special. It uses a type of statistical analysis to control the way it quantizes—and what this does is, it tends to decouple the quantization noise from the musical content. It lets you hear through the noise floor better, and gives a far more convincing sound picture and sense of realness. I think for many types of music this is by FAR the best wordlength reducer I’ve done.
The special monitoring modes are from freebie plugins, built into Ditherbox for quick reference and convenience.
SlewOnly gives you a calibrated version of just the slew of the audio, with all bass and most midrange stripped away. It’s an intolerably bright sound, which should generally not be coming across louder than the unaltered mix, but doing this to the mix should not destroy its internal balances too much. You can adjust balances of trebly content while monitoring through this.
SubsOnly is the same thing, for subsonic bass. On some speakers this output will be more or less inaudible. It’s calibrated like SlewOnly, so that its peak and RMS balances should come out about the same as the regular mix. If this turns out significantly hotter than the original mix, you can expect trouble on bassier systems. Like the SlewOnly mode, you can adjust balances of relative bass content listening through this, and return to the full mix to focus on more general issues.
Noise Silhouette monitoring has no notes: it is a burst of slightly darkened white noise with the dynamic shape of the mix. It can be used to check that you’re not losing rhythmic integrity, or burying parts that should carry part of the rhythmic load. You should be able to make out mix elements through their dynamic signatures, and hear drums and mix elements by their dynamic signatures. If you can’t make out any elements, the mix is overcompressed or overlimited. If certain elements are disappearing, they’ll also be lost in some playback situations. This will also clearly show excess low bass content that your monitoring could be missing—it will present subsonic information as a burst of hearable noise of equivalent volume, making it a possible tool for dialing in highpasses such as the ones in ToVinyl.
Ditherbox is $50.