By Jason Snell
June 19, 2015 10:53 AM PT
Podcast nerdery: Subtracting audio from a stereo file
Note: This story has not been updated for several years.
Here’s some podcast/audio nerdery that won’t be of interest to most people, but it’s saved my bacon more than once and just this morning it appears to have saved the bacon of a fellow podcaster, so here goes.
I broadcast my podcasts live using Nicecast, a $59 utility from Rogue Amoeba. One of Nicecast’s, er, nice features is that it’ll also optionally save an archive of your broadcast locally. I’ve enabled this feature, mostly just in case the recording software I usually use—Call Recorder—fails.
There are a lot of reasons I use Call Recorder, most specifically that it records whatever microphone is selected as an input in Skype, so if you sound okay to your fellow podcast participants, your recording will sound okay too. You won’t believe how many times I’ve had it happen that someone has sounded great on Skype, only to send me a local recording of themselves that was made not with their fancy USB microphone, but with the lousy microphone embedded in their laptop or with the (somewhat less lousy) microphone on their earbuds.
But a failure in Call Recorder can be catastrophic. Call Recorder saves its files as QuickTime movies, and if the program doesn’t finish saving that file—say, there’s a crash or a power failure—the entire thing is unsalvageable. So it’s good to have a backup, if not more than one.
Anyway, I had a recording failure a few weeks ago and turned to my Nicecast backup. When I opened the file, I discovered something curious: The file was a stereo recording with both my voice and voices on Skype on the left side, but only the voices of my panel on the right side. This probably happened because I’m using a stereo USB audio interface but only a single microphone.
So I had a thought. Having an isolated audio track of my own voice would improve the quality of the recording and reduce the amount of time I’d spend editing the podcast. Could I somehow subtract the content of the right side from the left, leaving me with a recording of just my own voice?
The answer turned out to be yes. I used my basic audio touch-up tool of choice, Sound Studio, to recover my microphone audio and save the day. First, I copied out each side of the stereo track into their own individual mono files. Then I selected the entire contents of the right track (the one containing just my panelists’ voices) and chose Audio: Invert Signal Polarity.
Go back to high school physics for a second. A wave can be cancelled out by an identical, but inverted wave. This works in the ocean (where two waves can interact and end up cancelling each other out) and it works in sound, too. It’s also a principle used in noise-cancelling headphones.
Anyway, once I inverted the polarity of the panelist-only signal, I copied the result and switched to the window containing the audio of my voice and the panelists together. Using Sound Studio’s Mix Paste command, I pasted the inverted sound over top of the original. And, much to my surprise, it actually worked! The mix paste had subtracted the other voices from the file, resulting in a track that contained only my voice.
Though the quality of the Nicecast archive wasn’t as high as my Call Recorder file (because I was using lower quality settings for the backup), it was still pretty good. I used the track in an episode of The Incomparable and I’m pretty sure nobody noticed a thing.
Like I said, I’m not sure how often this sort of thing comes up in the real world. But if you ever run into this problem, I hope you’ll remember this story and try this approach. It might save your bacon like it saved mine.
If you appreciate articles like this one, support us by becoming a Six Colors subscriber. Subscribers get access to an exclusive podcast, members-only stories, and a special community.