Pre Mastering Tips

Mastering is a completely, completely different animal, and I'm not sure how much I have to contribute to a general discussion of mastering other than to say don't master your own mixes. If they don't sound right, fix the mix. There is nothing that a home recordist can do in the way of self-mastering that they can't do better by re-mixing.

The point of “mastering” (in the sense that most recording forum-goers think of it) is to fix the stuff that's wrong with the mix. Which, if you can hear it in your own mix, should be fixed in the mix. That way the mastering engineer (even if it's you) doesn't have to worry about anything other than duplication. What nerd-factormax is doing is not actually “mastering.” It is “making a two-track mix sound better.” Which is something that mastering engineers frequently do, because they often get flawed mixes that they can improve, but that process is a sort of “pre- mastering” (in fact, pre-mastering is exactly what it is called).

I had a hunch that the Loudness Race strategies would have wider applications, but my specific scenario at the moment is this:

Trying my hand at mastering (for the first time) on a friends band. The tracking/mixing was (it seems) done in a bedroom studio and suffers from uneven bass guitar (1 or 2 notes dominate and the rest disappear) and massive transients on the kick and snare eat- ing up all the headroom.

The songs are destined for the myspace arena, so trying to retain quality is partly an academic exercise. I tried to get the average loudness somewhat close to a reference cd the friend gave me, but struggled to deal with the kick and snare transients.

Main Questions:

When dealing with transients and trying to gain headroom, is a compressor with pre- comp useful or should I turn to soft/hard clipping/limiting? AND should low frequency transients (kick) be dealt with in a different way? I tried twiddling with some precomp and quick release, but gained hardly any headroom before noticable pumping.

220From that point of view, audio is audio and there is nothing different about pro- cessing a two-track recording of a whole band than anything else in this thread. You can do whatever you want to make it sound better. Frankly running the whole thing through a distortion pedal should not be ruled out.

To the specific point of evening out the dynamics, you could obviously a com- pressor or a limiter, or use Smurf's method of manually drawing in envelope changes, or simply ride the faders. The obvious challenge with any of these is that whatever method you use to drop the level of the kick drum is also necessarily go- ing to drop the level of every other instrument. Whether that pumping is good or bad, and which method will be the least offensive on any particular source mater- ial is an open question, and the “best” answer depends on your skill, your gear, and how much time you want to spend.

Using a multiband compressor can mitigate some of the pumping artifacts. For ex- ample, if you took two tracks of the exact same source material, and use a high- pass filter to cut off all the lows at say 500Hz on one track, and used a lo-pass fil- ter to cut off all the highs ABOVE 500Hz on the other, and then sum the two tracks together, they'll basically be the same as the original source material. Ex- cept now you can apply a compressor to ONLY the material below 500Hz, and then mix the compressed lows back into the unprocessed mids and highs. This might allow you to sculpt the dynamics of the kick and low bass without causing the sensitive vocals to suck and pump in and out on the kick hits.

The above is the oldest and original form of multiband compression. But lots of modern multiband compression (a.k.a “dynamic eq”) plugins automate the whole splitting and summing business, making it very easy to have lots of bands, maybe including one “tuned” to the spikiest frequency of the snare drum or some such. So there's one approach.

Parallel processing is another potentially useful trick in situations where it's hard to get the just the right processing. You basically clone the track, then apply the hardest, flattest compression (or whatever) to one track, to try and completely flatten out the level, and then mix it back in with the unprocessed track, allowing you to more easily “tune” the critical balance. This is sometimes called “Motown” or “New York” compression (don't ask).

You could try multiband processing on the parallel, and you might get a very transparent form of compression. You could take it a step further and try filtering the “unprocessed” track with eq or some such to try and zero in on say the vocals and strings or some such to try and really separate the two streams into “needs compression” and “needs to sound uncompressed.” You might even end up with one compressed track of just the lows, one uncompressed track of the whole mix, one focused, uncompressed track of the “vocal focus” eq, and one compressed track of “snare focus”, then mixing those four stems together to effectively try and “remix” the song as though you were mixing a live multitrack with a lot of bleed.

You might also find that some reverb or delay applied after the compression (of whatever sort) might help to “smooth over” the pumping. You might also try some saturation or mild distortion effects (maybe in parallel or multiband) to substitute a little added “crunch” in place of compressor pumping. This might happen either in place of or in conjunction with compression, parallel, multiband, whatever.

A mastering engineer might also use creative phase cancellation, gating and eq'ing a clone to isolate just the kick, snare, and bass, and then inverting the phase and mixing with the original stem to try and reduce those specific elements.

Last but certainly not least, don't overlook plain-jane eq. If you can hi-pass at 50Hz, shelve down everything below ~12k by 4db, and scoop out some of the “mud” frequencies at ~250Hz, then you might be able to turn up the track by 4dB or more before clipping.

The slightly lighter bass might be less offensive than compression artifacts, and will almost certainly make the band happy in a straight A/B test if your ethics don't prevent you from “cheating” by using the loudness effect to deceive them in that way. You could also of course combine this version as yet another stem with some of the other processes above.

Good mastering engineers will combine any and all of these techniques, and oth- ers, to “remix” material as necessary in the premastering stage, which gives them a reputation for being magic-workers, which in turn leads to the misperception that “mastering” is the key to great sound. But as you can see from all the above, it would be much, much easier for a home recordist to simply go nac and re-work the mix if the kick and snare are too loud.

These kinds of techniques push the technical and aesthetic limits of audio pro- cessing, and working all these stems and crossover filters for multiband and so on tends to introduce progressively more and more phase smear and other pro- cessing artifacts. For this reason, professional mastering engineers tend to be pretty obsessive about using specialized, high-quality equipment and minimalist processing whenever possible. With 50 stems and unlimited processors, you could practically remix the whole song this way, but the audio degradation would be worse than the improved mix. Moreover it would be exponentially more difficult and time-consuming than simply doing a remix of the original source tracks.

"High-pass" means the filter "passes" everything above the cutoff frequency and blocks everything below the cutoff frequency. So the filter shape should sharply curve down to an infinite gain reduction, depending on the steepness of the filter (Q setting). “Low-pass” is the same thing, except in reverse – the highs are blocked, and lows are “passed” through the filter.

A “shelving” filter is just like a “step” or “shelf” that evenly raises or lowers everything past the filter frequency. It's like a tone control on a stereo.

As an aside, using terms like “high pass” to refer to an eq curve that cuts all the lows might seem a little confusing – why not just say “low cut”? The thing is that “low cut” could refer to a shelving filter or even a notch filter. “High pass” clearly describes a hard cutoff filter that only passes frequencies above the cutoff.

Hope that makes sense.

The pre-ceeding post was some compiled opinions and excerpts from threads written by some great folks on the Reaper forum, the moderator mentioned their was no problem with us creating our own version with the information, so the following is some of the best information and concepts that i believe you, my readers will find especially helpful. I couldn’t have said it any better, these concepts and ideas are top notch, and pure gold. You can read the entire 53 page massive thread if you’d like, here. http://forum.cockos.com/showthread.php?t=29283