A few Homerecording.com regulars debate the merits of dithering. The conversation could easily have devolved into a flame war, but the participants kept it civil, and offered some great food for thought.
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?
That quote comes from Ethan Winer’s great summary of his position on the matter – he’s squarely in the “dithering is usually not needed” camp.
I tend to agree with Ethan. Responsible mixing engineers don’t apply processing to a mix if they themselves don’t hear the effect of the processing. Simply put, if you can’t hear a difference, don’t make the change.
Unmitigated awesome: Daved Lee Roth’s vocal track from Runnin With The Devil, solo’d.
Converting Ikea bedside tables into studio racks: “the Rast bedside table makes a snug rack for music machines.”
Two unrelated sites feature famous songwriters discussing what went on behind the scenes as they wrote:
First, Joni Mitchell on the writing and recording of her most recent album:
When I recorded it, I was sick so a doctor prescribed some penicillin, which I had an allergic reaction to. I was delirious, stressed out, and we worked all night long. I was so delirious that I was playing way back on the beat… [I]n January 2007, I had demos of the Shine songs with me and played them to some friends at a party afterward. James Taylor told me that he had to play on this song. I wasn’t sure if anyone could because it was created in such a rare spirit. But James came in anyway and I asked him to play short figures like a saxophone. So you can hear fractions of James’ guitar playing here.
Jim Vallance’s site has some fantastic insight into the mind of a professional songwriter. Jim, who’s worked with Aerosmith, Ozzy, Bryan Adams, The Scorpions, and Thornley, meticulously lists every song he has ever written. The site is full of anecdotes and details about his creation process.
On our very first basement demo of “Summer of ’69” we started the song with the 12-string riff, exactly like the “break down” section in the middle of the song … but on subsequent demo’s we replaced the 12-string with a chunky 6-string intro. In fact, we toiled over the musical arrangement for several weeks, maybe longer. We recorded the song three or four different ways, and we still weren’t convinced we had it right! Bryan even considered dropping the song from the Reckless album.
Now, 20 years later, when I hear “Summer of ’69” on the radio, I honestly can’t remember what bothered us.
Tags: arrangement, hearing, myths, vocals
8 comments
Trackback URI Comments feed for this article
“If you can’t hear a difference, don’t make the change.”
Key point. I have heard a lot of talking on musical issues, that forget about listening. To talk about the quality of algorithms (dithering, compression, etc.) is easy. To illustrate the talking by having people actually listen and hear the differences may not be that easy… and is done far less than talking.
Responsible mixing engineers don’t apply processing to a mix if they themselves don’t hear the effect of the processing.
But dither isn’t processing, it’s part of the definition of correct bit-rate reduction. *Lack* of dither is processing–you’re adding signal-correlated distortion by deviating from standard practice.
Not dithering is like playing back a Dolby-encoded tape without Dolby decoding because you want to avoid processing. You’re doing the reverse of what you mean to do.
Does it matter? I’ve been able to hear the difference in some circumstances–the undithered signal lost stereo image. It’s subtle, but precisely because it’s subtle, you might miss the fact that you’re doing damage by not dithering.
Good point Tim.
I harp on a few things around here, one of them being the importance of using your ears to tweak a mix, rather than relying on formulas and general rules. I was probably reaching a little trying to work it in up there. (Though I’d say my advice stands, even if I used it out of context.)
> because it’s subtle, you might miss the fact that you’re doing damage by not dithering.
That cuts both ways, of course. :-)
That cuts both ways, of course. :-)
When you buy a new piece of gear, do you open it up and start pulling out capacitors as long as you don’t hear any difference?
Because that’s what not dithering is like. A bit-reducer that doesn’t dither is broken, pure and simple. The brokenness may not be audible (to a specific person on a specific monitoring system at a specific time…), but you still shouldn’t do something you know is broken*, especially bearing in mind that you don’t know what use might be made of that audio in the future–it might be something that will expose the problem.
Sorry to be such a pedant, but… I am a pedant. :)
*Except as a deliberate effect, of course.
> When you buy a new piece of gear, do you open it up and start pulling out capacitors as long as you don’t hear any difference?
Does this count? http://www.hometracked.com/2007/08/17/ribbon-mic-mod-part-1-apex-205/
:)
> but you still shouldn’t do something you know is broken
Ya, I completely agree. And I’m glad you pointed out my misstatement. Like I said, I was trying to shoe-horn a point into the wrong context.
> but… I am a pedant. :)
heh, it comes with the territory, I think. So much in sound engineering is about the details!
Thanks for underststanding! I’m enjoying the site quite a bit, by the way, and not finding anything else to quibble about…
I should add that I agree with your main point, that it’s better to sound good than to be “right.”
I used the built in ditherer in Cakewalk pro audio 9.3 last week, (to change down from 24 to 16) and found, with a fairly hot mix, that there were quite a few bits of click/pop sort of artifacts. I assume the algorithm doesn’t like hot mixes or that there was conflict with a plug.
I’ve yet to try a plug in ditherer. Are there any recommendations?