Most of our recording gear now records in 24 bits. A 24-bit “word” looks a lot like 110101101000110110011010. In base-10 this can represent any number between 0 and 16777215. Here is the catch. 16777215 doesn’t represent a LOUDER sound. We are not adding “dynamic range” when we record in 24 bits. We are actually adding more resolution. Think about it in terms of the page you are reading right now. If we think of this page as being a finite size, by increasing the dots per inch of the printing, we see a higher resolution; a more realistic page. That is the difference between 24 and 16 bits.
The problem with recording in 24 bits and then reducing the recording to 16 bits isn’t a change in dynamic range (or loudness), it is how best to reduce the resolution from very good to good. If you have a nice digital camera and take pride in your pictures, there is usually a disappointment when you click the “e-mail my picture” button. The result is often good, but sometimes that reduction in size and resolution makes your picture look grainy or jagged. This is VERY similar to the bit-depth problem found in reducing 24-bit recordings to 16-bit. Now just like certain pictures don’t reduce well and others do, some audio reduces well while other audio doesn’t. It gets worse! The audio that doesn’t reduce well is the audio that looks “spiky.” It is the sharp transients – drums that suffer most; especially cymbals! If you think about audio as being like a “dot to dot” drawing, when you have a whole bunch of tightly spaced dots (like there are in 24-bit audio) when you take dots out periodically (to reduce the data as you would going from 24 bits to 16 bits) how do you take the right ones out in order to keep the same shape? This is the “big deal” about dithering.
How do you choose a dithering option? Well this is where (as an engineer) you get to be like a wine connoisseur. You know that “uncomfortable” moment you can feel at a fancy restaurant when you are presented with the wine and you have to “taste” it properly? This is not all that dissimilar. You can actually use your ears to decide on dithering provided you know what to listen for. Don’t listen to the bass; you can listen to the top end, the little tweeters do respond quick enough to actually show a difference in resolution, but the real place you are going to “hear” dithering (if at all) is going to be found in the punchiness. No, it is not necessarily going to make your mix “punchier,” rather, dithering will show up as maintaining the realism of the snare or cymbals. The “crack” of a snare drum is simply more realistic in 24 bits than it is when captured by 16 bits. Once you catch on to the difference, it will become fairly obvious which choice you should make. If you’re recording acoustic drums or instruments, dithering can be quite important. I will end this by saying, as usual, that these differences are pretty small; the biggest differences you can make in a recording are by getting good players to give you good tones. The most impressive recordings capture performances. The correct dithering can maximize the tonal quality of the performance, but it is not going to fix (at all) a poor sounding recording. A crappy 24-bit 192 kHz recording is still crap … it’s just high-resolution crap!
Alec Watson is a Producer/Engineer that works from his destination studio on Vancouver Island.