Last week, we started our podcast production series taking about the first step--recording. If you missed it go back and read it here. We'll wait. Today, we're going to talk about post-production.
The first step in post-production is to choose your editing platform. Here, you have dozens of options. Logic, ProTools, Cubase, Nuendo, Audacity, Audition; some free, some expensive. Often times, when you buy a USB or FireWire interface, you'll get a recording/editing app for free (though sometimes it's a light version). They all do the same thing, and the differences aren't so much about quality as they are the user interface and feature set. Unless you're not doing a good job getting a quality recording, the need for extensive editing features is minimal. What you want is a clean, fast interface and a good compressor plug-in (this is assuming you're producing a sermon podcast--if you're doing the whole song set, the rules change, but that's another post).
I use Soundtrack Pro for a few reasons. It came with FinalCut Pro Studio, and I like the clean interface. I also really like the Platinum compressor. It does a great job without sounding like a compressor.
For a typical weekend message, I really only do three things in Soundtrack; edit heads and tails, normalize, and compress. Occasionally, I'll do some editing if there is a good reason to, but typically we post the whole message. Let's look at those steps one at a time.
Heads and Tails
This is pretty easy. Find the beginning of the message, back up a second or two and put a fade up on the clip. Something that drives me nuts is hearing podcasts with 5-8 seconds of silence at the beginning. Limit it to a second or so. A 1/2-1 second fade in is really nice. Same goes with the end of the message. If the band is underscoring, I'll look for a musical place to fade out so it doesn't sound abrupt. I'm not anal about this, though, I figure people are listening to hear the message. When it's over, they're done.
In my recording structure, this is critical. I always leave a good 12-15 dB of headroom when I record. That is, my signal level never gets above -12 dB during the message. Remember, this is digital recording; once you run out of bits, you're done. Distortion is quick and nasty. There's no reason to try to maximize headroom at the recording stage. Give yourself some safety margin. If the preacher suddenly gets excited and starts shouting (which has happened more than once), it's good to know I've got plenty of margin on the recording.
Recording this way means that I need to bring the levels up closer to 0 in order to maximize the end-user listening experience. You don't want users to have to turn their volume all the way up every time they listen to your podcast. That's irritating. So we normalize.
Normalization will bring the peak level of the entire clip up to a level you specify. It's kind of like turning up the volume, only smarter. A lot of people like to normalize to 0 dB. I set it to -2 dB. I like to leave a few bits there so we don't saturate in the next step. Normalizing typically means selecting the clip, choosing Normalize from a process or effects menu and specifying your peak level.
You may have heard me decry the use of compression on music in the past. I can't stand overly-compressed music that has had all the dynamics sucked out of it. However, when it comes to a sermon podcast, that's exactly what I do. Well, almost. Here's why: People listen to sermon podcasts in noisy environments--at the gym, in the car, on a walk, at the computer--and they don't want to be turning the volume up and down to keep the level at an even keel. It's just annoying. So I compress the sermon pretty hard to keep the perceived level much more consistent.
As I mentioned, I like the Platinum compressor. Here is a snapshot of my typical settings:
My typicall settings for my sermon compressor--saved as a preset.Note a few things: First, the threshold is pretty low (-20 dB)--that keeps the compressor running most of the time. In fact, it's typically running at 6-8 dB of gain reduction, and when the speaker gets going, anywhere from 12-20 dB. I count on the timber of the speaker's voice to give dynamic cues. I also have auto gain turned on. This is set to 0 dB, and tries to bring the level up to somewhere approaching 0 dB. Then I turn on a limiter to limit to -2 dB. The effect is a very consistent perceived level. However, due to the tonal cues in the speaker's voice, you can hear when he's louder of softer. Tricking the ear is fun!
Also note that the attack and release thresholds are pretty fast. I want it to jump in and clamp down quickly, but release quickly as well. My ratio is a mild 2.5:1, which keeps it from pumping or breathing too much. It sounds quite natural, really.
The final step, though not really a step, is to export a stereo AIFF file at 44.1 KHz. I could export to an MP3 from Soundtrack, but I don't think the built-in MP3 encoder is good enough. For that we need to employ another tool. And that will be the topic of next week's Audio Monday post.