Recording - November 2015 - 61

recorded tracks are being mixed down. Compression that reacts
before the fact is called look-ahead compression.
It's common to use compression on a bass guitar track, to compensate for a player's less-than-even touch (and, sometimes, the unequal
outputs of the instrument's strings). But if you compress the signal from
a bass instrument using both fast attack and release times, it will produce rather gross-sounding distortion, soft-clipping the waveform. The
way to solve this is by increasing the compressor's release time.
Compression in our sample session
For the Ottoman Underground session (audio at
http://is.gd/RECFirstStepsClip1), I'd planned to use some light
compression on Shlomo's bass tracks. As it turned out, though,
I didn't need it; his touch is so even, and his instrument so wellbalanced, that I used his tracks with no processing. As you can
hear from the clip, it sounds fine.
If I'd needed to compress him as we recorded, however, I'd
have combined the miked and DI bass tracks in a submaster mix
on the board, run the resultant submaster into a compressor, and
sent the submaster mix to the multitrack recorder as well as to
the live-to-2-track mix.
What I did compress was Will's vocal. I mentioned before

Note that I did the compression as we tracked; it was recorded
on the multitrack rather than added in the mixing stage. Why? I'm
old-school enough to have developed the habit of doing compression on the way in; in the analog days it avoided the possibility of saturating the multitrack tape, and with digital it makes
digital overloads (which sound a lot worse than analog) unlikely.
In addition, I processed the signal going in because in addition to the multitrack, I was also making a direct-to-2-track
recording of the song. I did that because I wanted to give it to
the musicians fairly quickly, so they could critique their performances and decide which takes they wanted to use.
I didn't use compression on the banjo or clarinet. The clarinet's level stayed nice and consistent (Dana's classically
trained), while the banjo lost its spiky emphasis with every compressor I tried. And I wanted that spikiness-it contributes to the
song's drive and energy.
Conclusion
I've barely scratched the surface of compression. I haven't
talked about using compressors on drum kits; that's well worth
an article in its own right-but hey, there wasn't a drum kit on
this session. I also haven't touched on parallel compression,

Figure 2: Feedforward Compressor

Figure 3: Feedback Compressor

that compression can make a vocal sit better in a dense mix.
Now, this song is the antithesis of a dense mix, but it turned out
that some very sparing compression did wonders for the vocal
and its blend with the other instruments. On this song, possibly
in reaction to the sharpness of the tenor banjo sound, Will tended to "point" his vocal, emphasizing certain words. That sounds
fine in live performance, but smoothing it a bit with a compressor worked better for the recording.
As it happens, the console I was using for the session, an SSL
in my university's main recording studio, actually has simple
compression circuits built into each channel strip. These use voltage-controlled amplifiers (VCAs), and they provide a nice way
to get unobtrusive compression. I set the ratio at 4:1 (my usual
starting point for vocal compression), but I set the threshold to a
point where the compression was being triggered only on Will's
loudest vocal peaks, where he was adding sharp emphasis to
his voice. As a result, most of the time the compressor wasn't
actually doing anything; it only acted on the loudest notes. I
used a fast attack time and a medium release time, which I've
found works well for this sort of minimal vocal compression.

multiband compression, or deliberately using compression artifacts for creative effect.
After some discussion, our Editor and I drafted a short sidebar
on expanders and gates, which work rather like compressors in
reverse and can be used to suppress background noise and control mic leakage-albeit with some tricky sonic tradeoffs. But
these devices, which actually increase dynamic range, can be
hard for a novice to understand (and aren't nearly as commonly
used nowadays as they were in the analog era), so we elected
to set that topic aside as well. After all, this series is about First
Steps, and like so many topics in recording, covering everything
involved in dynamics control would require a book rather than
an article.
Now it's time to talk about the other type of signal processing
I applied while tracking the session: equalization, or EQ. See
you next time!
Paul J. Stamler (stamler@recordingmag.com) is a recording
musician, engineer, educator, and collector of bizarre vintage
recordings, living in St. Louis.
RECORDING November 2015

http://www.is.gd/RECFirstStepsClip1

Table of Contents for the Digital Edition of Recording - November 2015