every day is another shooter producing a hugely successful compression programs comhtml
I'm about to settle down to edit this week's podcast, and I'm going to tell you all how I do it. Well, the audio side of things, anyway. Not so much as a flawless how-to guide, but more a SHITTY STICK in the collective eye of those who think it's just a case of hitting a record button, chatting on Skype for an hour, then uploading the results to the internet. I dare say that's how some other, more 'professional' sites might do it, but that ain't us.
I'll mainly aim this at people who know very little about audio editing, so apologies to those of you who do.
Step 1: Recording
We all talk on Skype, as Andi lives a good 180 miles or so from me and Ian. Now, while there are loads of programs available for recording Skype conversations, we choose to each record our individual direct mic feeds. This means that instead of having one, Skype-compressed audio file* with all three voices on it, I have three audio files, one for each participant, which sounds a fair bit clearer and makes editing a lot easier in a number of ways which I'll describe later. It also means we're not at the mercy of any connectivity issues ¨C if one person's internet connection goes a bit funny, their voice will distort, and a Skype recorder would record that. But if you record the mic directly, it makes no difference. On top of all that, we start each recording by counting to three in unison, to make it easier to sync up the recordings later.
If anyone cares, we all use this headset. It's served us well, and the only complaint I could make is that it's a bit too sensitive, and tends to pick up noises other than the user's voice. But, more on that in a minute. Oh, and we all record with Audacity ¨C it ain't clean, but it's cheap. Free, in fact.
Step 2: Editing
Once Ian and Andi have sent me their individual recordings, I can get on with the editing. I use Cubase SX, not necessarily because it's the best, but because I was taught to use it in college and have seen no reason to change. So, suck it.
Over the months I've settled on a standard setup (helped by the fact that we now all use the same microphone), which has made things a lot bloody quicker. Once I've imported the audio files, it looks a bit like this:
So that's the three recordings, each on their own track, with the top track being reserved for the intro/break music. Now, if we were lazy, we could just cut out the bits we didn't want (i.e. the bits at the start before the show kicks off proper, and the toilet break in the middle) and export that. But we can do a bit better than that, mm?
Since we all use the same mic headset, I only need one set of parameters which I can then copy and paste to the other tracks ¨C and now that I've got a setup I'm happy with, it's saved and ready to go each week. I had to make a few minor alterations after copying the settings to all the tracks ¨C Andi's audio tends to be a bit louder than the others, for example ¨C but they're basically the same. Here's the config windows for all the processing I use on my track:
But what the fuck is that all for, eh?
First off, I mentioned earlier that our headsets are a bit too sensitive ¨C the main complication here, aside from it picking up unwanted noises from our bedrooms like PC fans, mouse-clicks, and so on, is the fact that it picks up the output from the headphones. So, when one person is totally silent, you can still hear the others talking ¨C albeit very quietly ¨C on his audio feed. And that needs to get filtered out. So, I use a noise gate ¨C in this case, a free one called Floorfish from the excellent fish fillets package. This means that when the volume on the track falls below a certain point, the audio is cut out completely. And since my normal speaking voice is a lot louder than the feedback from the headphones, it's easy for the noise gate to distinguish between the two.
The second one, Blockfish (from the same package), is a compressor. You'll remember I mentioned the distinction between data and dynamic compression ¨C well, this is the latter. Dynamic compression is where you 'squash' an audio sample so that, in this case, whenever someone speaks, it's more-or-less always at the same volume, whereas the volume of natural speech tends to change a lot. This gets used a lot on radio stations and pop music, because it makes sure things can be heard when you're listening on your shitty car stereo. Some radio producers don't like it because, by its very nature, it will rob recordings of their subtlety; but we don't really do subtlety on our podcast, so we use shiteloads of compression.
Another problem with dynamic compression, though, is that it can introduce excess sibilance. Basically, that means it sounds fucking annoying when people use 'ess' or 'shh' sounds. To fix that, we use a De-Esser, which does pretty much exactly that you'd think, and stops us killing people's ears.
The bendy line in the middle is the equalisation ¨C what you can see there is that I've pulled out some of the lower frequencies, as the headset we use sometimes produces a very low hum. It doesn't always do it, but a low-frequency cut like that has very little effect on our voices, so it stays in, just in case.
What comes next sort of depends on how busy I am. If I haven't got an abundance of free time, I can scan through the whole thing, delete the 'break' bits, and get Cubase to export the MP3. But if I'm not in a hurry, I'll listen to the whole thing and tidy up any awkward bits where, say, I laugh too much at one of my own jokes, or two of us talk over each other ¨C having each participant on a separate track means I can just pull bits out from one track without messing with the others.
As a final tweak, I do a bit of stereo panning ¨C I put myself over to the left speaker, Ian in the middle, and Andi on the right. People often accuse us of all sounding the same, so this helps them to differentiate between us, since the voices are coming from different directions.
And that, my darlings, is basically it. Until I do the follow-up post to this, where I'll talk about the process of actually tagging and publishing the audio file. It'll be a bloody riot, I promise you.
Ta-ra! x
*(that's data compression, not dynamic compression. I'll explain the latter further on, but in this case we mean an audio file with loads of information taken out, to make it smaller).