- Coffee Shop DevOps: How to use feedback loops to get smarter
- My previous podcast with Jen on distributed teams
- DevOps at Red Hat overview
MP3 audio (14:55)
OGG audio (14:55)
After something of a hiatus, my Cloudy Chat podcast is once again publishing. (iTunes feed. It’s also now on Google Play and available directly from my blog.) My regular co-host had moved onto new responsibilities at Red Hat and with a lot of travel and other things going on, I just let podcasting lapse. But I’ve got a new episode out with Red Hat’s Jen Krieger talking distributed teams, another one mostly in the can, and am partway through refreshing some of the standard audio and graphics associated with the podcast. I also expect to experiment with the format a bit going forward so you’ll likely see (well, hear) some approaches that differ from my standard 15-30 minute straight interview.
With this relaunch, I’ve gotten some questions from colleagues asking how I go about recording and editing podcasts. I’ve written on this topic before, but reading back, I see that I’ve made a fair number of changes to how I do things. So time for an update.
Let me note at this point that I use a number of different approaches depending upon the circumstances.
I’ve written about how I record an interview with a remote guest previously and that description still pretty much applies except that I now generally use BlueJeans rather than Google+. But it really doesn’t matter; the process should be pretty much the same for most video conferencing systems.
If I’m on the road, I try to minimize gear and use one of the iRig microphones plugged into my iPhone. If I recorded this way a lot, I might invest in a dedicated recorder. Note though that some of the high quality ones don’t work well as hand-held recorders because they’re overly sensitive to being handled while recording. (The TASCAM that I describe later on has this property; you have to mount it and/or use external microphones.)
For this post, I’m going to describe how I record interviews when I’m in the office and don’t mind bringing in a (small) bag of recording gear. So this describes a case when your co-host(s) or guests are often in the same location as you are. I’ve fiddled with my kit over the past couple years and I’ve settled on this setup as one that is pretty straightforward once you have it down, can give excellent audio quality, and works to create a natural-feeling interview environment.
My hardware is as follows:
Setup the recorder for the external mics. I also use auto levels to try to balance the volume of the microphones but it doesn’t work as well as I would like. But, with my configuration, the mics record on different stereo channels so I can do some manipulation before I blend them. (I also have a USB sound mixing console that I can use to attach more microphones and to directly control their volumes but I’ve found, for most purposes, this adds a lot of complexity without a lot of benefit.)
For editing software, I use Audacity which is Free and Open Source and has far more features than I need or use. It also runs on Linux, Macs, and Windows.
Once I’ve blended the channels, the first thing I usually do when editing is go to Noise Removal under Effect, get a noise profile, and then apply noise removal to the entire recording. I think this capability was introduced in version 2 and, in my experience, does a great job of removing ventilation and other consistent background noises. Of course, the quieter an environment you can find, the better.
After editing in Audacity, I prepare the XML file that’s needed for iTunes and Google Play, splice my standard intro and outro audio onto the beginning and ends of the podcast, and upload the XML and the audio files to AWS. Most of the podcast apps get their feeds from the iTunes API these days, so that’s the one upload that really matters. Even with Google Play, you need to go through the steps to get your podcast into their store, but after that it can draw from the same XML file that iTunes uses. The only separate upload I do is to Soundcloud.
If it sounds kinda fiddly, it is. I’ve written a Python script  to take care of the splicing, the XML creation, and the AWS uploading. (Basically, I fill out a form with the title and description, point it at the MP3 file, and it does the rest. The only manual steps I still have to do are uploading to Soundcloud and writing a blog post for the episode; I also typically get my episodes transcribed by CastingWords for inclusion in the post.)
 It occurs to me that this is probably a good use case for Amazon Lambda but I wrote the original version of the script a few years ago and it generally works fine.