My podcasts are almost invariably recordings of phone calls. Following the advice of my audio guru, Doug Kaye, I’m using a Telos ONE to achieve decent audio quality using POTS (plain old telephone service). But from time to time I revisit the question of whether Internet calling, using Skype or another voice-over-IP solution, can produce results of equal (or better) quality.

The good news, since the last time I tried this, is that it’s easier to record Skype calls. The recipes used to involve a whole lot of baling wire and black magic. But now there are Skype add-ins that simplify things quite a bit.

For today’s test I used two different products: MX Skype Recorder for Windows, and the Ecamm Call Recorder for the Mac. Because friends and family now refuse my requests to involve them in audio recording experiments — and who can blame them? — I manned both ends of a Skype call, shuttling between a PC in one room (with an analog headset) and a Mac in the other (with a USB headset).

Doug Kaye’s recipe for using the Telos ONE involves splitting the caller and callee onto separate channels of a stereo recording. That enables the kind of editing I illustrated in this brief screencast. I’m happy to report that both of these Skype recorders enable the same kind of thing. MX Skype Recorder will directly produce a split-channel WAV file. The Ecamm Call Recorder produces a QuickTime movie with two stereo tracks, one for each half of the call, but you can extract them and recombine the parts to achieve the same result.

So that’s all good. But when I finished my test call and reviewed both recordings, I found in both cases that while the quality was fine for the local voice, it was sketchy for the remote voice. In particular, listening to the recordings made from each end of the call, I hear the occasional dropouts and compression artifacts that I always hear in every Internet call, whether it’s on Vonage or Skype or iChat or Windows Messenger.

Just for kicks, I took the two recordings apart, swapped channels, and put them back together to create two new versions of the test call. One combines both local voices and it sounds like this. The other combines both remote voices and it sounds like this.

I’ll be recording a podcast tomorrow and, since it’ll be an international call, I’d like to be able to use voice-over-IP. Based on these results, though, my conclusion is that combining two locally-made recordings — one of which the interview will upload and I will download after the call — will yield the best outcome.

In this case my interviewee is willing to play along, so we’ll give it a shot and see how it goes. In general, of course, that isn’t something you can expect an interviewee to do. But I don’t see a workable alternative. Skype-based podcasting still doesn’t feel like a first class option. Am I missing something?