Levels Management

One of the primary goals of mastering, as driven by client preference, is to make the CD as LOUD as possible. We are all constrained by the limiting factor of 0 dBFS in digital recording. Because louder, in a general way, sounds better (I’m not going to get into why this is so), the louder of two otherwise identical recordings will be preferred by end-users. The engineer that can most effectively push the level right up near 0 dBFS wins!

There are a couple of ways to do this.

In Each Track

We carefully compress each track, adjusting its threshold, ratio, attack and release to give us maximum musical and dramatic impact. We may also employ a hard limiter with a threshold at 0 dBFS as well. We don’t just want the music loud, we want it dramatically loud, and our compression efforts work on this. Often, multiple passes of fairly gentle compression will be used to build up an effective compression scheme.

Sometimes, too, we venture into the realm of overshoots (“overs”), usually at clients’ requests, increasing the level until they begin to complain about the distortion and then backing off. Not a pretty practice, but a fairly common one. Client is always right, right?

Along this same line is a hot new technique called “bit-shaving.” By stripping away several of the Least Significant Bits, we bring the loudness of the overall recording up. An interesting way to do business, don’t you think? 12-bit lives!!!

Normalization

Once we have the tracks compressed to our satisfaction, we normalize them, which is to say, we determine the maximum level of each track and set that at 0 dBFS. Now, each song on the CD is referenced to 0 dBFS, and the overall level of the CD is as loud as we dare make it. Needless to say, if we have crammed a bunch of overs into some songs to make them louder, we don’t then normalize their level downward.

However, just because we’ve normalized everything to 0 dBFS , it doesn’t follow that the levels of those tracks are musically matched. Normalization levels may need to be shifted to take care of such problems.

Corrective EQ

There are two issues in regard to equalization. The first has to do with equalization to correct for perceived flaws due to the effect of the control room monitors, such as when the low frequencies are deficient because the mix monitors were bass-heavy. Mastering engineers are really sharp when it comes to this.

The second equalization issue has to do with song-to-song consistency. Often, different tunes were mixed at different times, by different people, in different rooms. The mastering engineer uses equalization to reduce these differences, to bring the CD to a more focused and consistent sonic quality. They can really tease exciting mixes to emerge from somewhat marginal tapes by hearing out the frequency problems and fixing them. This is done very carefully, working to avoid the unfortunate possibility that by making the tracks more consistent we also lose the power and magic of individual mixes. Sometimes, little tweaks are added in solos, minor events like snare hits, etc., just to bring out the essential quality of the mix.

Management of the Digital Signal

Mastering engineers get DATs, hard disk files, analog tapes and occasionally other media from which to master. If the source recordings are analog, conversion is usually handled by outboard converters at 20 or 24-bit resolution, at the sampling frequency of the CD, 44.1 kHz.

If the source recordings are digital, the mastering engineer is stuck with the bit resolution and dither provided by the client. If it is at more than 18 bits, and at the right sampling frequency, the mastering engineer is happy, because he or she has some room in which to work.

To Truncate or to Re-dither?

Once all of the changes have been decided upon, and the final master tape is ready to be made, the master recording must be reduced to 16 bits for CD release. Usually, this will involve throwing away the unused bits and adding dither (from one of a variety of flavors) at the sixteenth bit.

Another possibility is to simply truncate the recording, throwing away all bits below the 16th bit, and not re-dithering. While this is generally frowned upon, one mastering engineer allowed to me that he does it a fair amount, to suit his clients’ occasional preference for that particular sound quality. Truncated recordings, he says, seem to have a little harder, edgier quality that some prefer, while dither tends to make low-level stuff a little rounder, smoother and sweeter. This has a family resemblance, of course, to bit shaving.

Finding Errors

With Sony 1630 recordings, we get a Digital Tape Analysis (DTA) that tells us about overs, mutes, holds, averages and parity errors. Finding errors on DAT or CD-Rs is much more difficult, and is currently generally done mostly by ear (which frightens the bejesus out of many mastering engineers!).

Getting An Acceptable Master

Usually, mastering engineers customize their work to suit the chosen CD factory (do they take 1630 format masters, for instance). Also, it is essential to create a safety AND a reference, so that while the client is impressing his/her friends, the master is not at risk. It is also essential to maintain good communications and relations with the CD factory. Most such factories are very open to this communication and co-operative effort.

Interestingly, there are only about 15 actual CD factories in the U.S. All other facilities offering CD production services are actually functioning as brokers. It is essential for the mastering facility (and the client) to know exactly who it is dealing with. Through a broker, obviously, the price of quality is eternal vigilance.

Post a Comment