Conventional audio wisdom says that dither is required whenever you reduce the bit-depth of an audio file. Typically, when reducing a 24-bit mix to 16 bits for putting onto a CD. Dither is a very low-level noise that's added when reducing bit depth, and by definition it's about 90 dB below the music when reducing to 16 bits. Most people would have a hard time hearing noise that's 60 dB below the music, since the music masks the noise. Yet if you ask a dozen audio recording engineers if dither is necessary when going from 24 to 16 bits, every one of them will say Yes. Even the manual for Sony Sound Forge claims dither is important:

If you want to burn a 24-bit audio file to an audio CD,
dithering will produce a cleaner signal than a simple
bit-depth conversion.

Some engineers even argue over which type of dither is best, claiming this algorithm is more airy sounding that that one, and so forth. But just because everyone believes this, does that make it true?

To test this for myself - and for you - I created a composite test file that switches back and forth several times between truncated and dithered versions, and made it available HERE (13.5 MB). These are the exact steps I followed:

I rendered a portion of a tune from Sonar at 24 bits. Some people claim that the benefits of dither are most obvious on delicate passages, so I picked a spot that starts with a soft gentle solo guitar, then goes into a louder rock part. I loaded the 24-bit file into Sound Forge and converted it to 16 bits by truncating, then saved it as a new file. I loaded the 24-bit file again and converted it to 16 bits with dither, and saved that as another new file. Then I pasted portions of the truncated file into the dithered file, and saved the result under yet another name and uploaded it to the link above. Sound Forge offers a number of dither options, so I followed the manual's suggestion to use High-pass Triangular with Noise Shaping.

I won't reveal here how many times I switched back and forth (it was several), nor where - you have to email me for the answer. I will say that all the splices were more or less on a beat. I made notes of where each truncated part was inserted as mm:ss.hh for minutes, seconds, and decimal hundredths. So if you email me which fragment you think is which, report in a similar format:

0:00.00 - 0:03.45 = truncated
0:03.45 - 0:15.22 = dithered
0:15.22 - 0:22.13 = truncated

Note that I did not make any edits at musical section changes, and no fragment is shorter than 7 seconds. All the splices are on musical beats or pickups, and some are right in the middle of a phrase. This exposes the differences the most (assuming you can identify them), versus switching when a guitar solo starts which would change the sound due to musical rather than technical reasons.

Added January 19, 2008: A forum friend and I did some extensive listening tests to compare dither, jitter, and A/D converter quality. My report is HERE in Lynn Fuston's 3dB audio forum.

Ethan Winer has been an audio pro and skeptic for most of his adult life. He now heads up RealTraps, where he designs acoustic treatment products for recording studios and home listening rooms.

Entire contents Copyright © 2007-2008 by Ethan Winer. All rights reserved.