Podcasting workflow

Here is how we typically record and edit episodes of Limitless Possibility.

Recording

We use the double-ender technique for the best sound quality. Double-enders are when both participants on a podcast record their audio locally for the highest fidelity. You also record a reference recording of the call itself, and during editing, you line up your local audio recording with the reference track, then mute the reference track.

It is recommended that you coordinate with your cohost(s) to use the same codec and sample rate to facilitate syncing of the audio later. If you use inconsistent settings, it may require additional effort to keep the audio lined up over the course of a whole episode, especially if the recording goes on for a long time (over 90 minutes).

The software you use to make the call itself doesn't really matter. We use LINE, record the reference recording of the LINE audio with Piezo, and record our local audio with QuickTime Player. Currently, we both have Blue Yeti microphones.

Audio processing

I first convert all of our lossy recordings into WAV files with Fission so they can be imported into Audacity without issue.

The reference recording from Piezo puts your audio on the left channel and the other call participants' audio on the right channel. I open the reference recording in Audacity, use the Split Stereo to Mono feature, and then export each channel as a separate WAV. I call these the dirty files.

Then I open up each of our local recordings in Audacity, turn the stereo audio into mono, and run the following operations:

Editing

I edit the podcast in Ferrite on the iPad, as the Apple Pencil support makes it one of the most time-effective ways to edit our show. My Apple Pencil drag shortcut is set to "delete", which means anything I draw over with my Apple Pencil will be deleted.

A typical episode will have 4 tracks: Yanik Dirty, Yanik Clean, Luc-Olivier Dirty, Luc-Olivier Clean.

First, I line up the audio from the clean track with the dirty tracks. Piezo only records while a call is active, so if a call disconnects, there will be sections that are only present on the local recording until the call resumes, and may cause additional lining up work.

Once the audio is lined up, I mute the dirty tracks. I prefer not to delete the tracks entirely until the show is completely edited, because occasionally I realize I forgot a step from the previous audio processing and need to line up someone's audio track again during the editing process.

I chop off any banter before the proper start and stop of the episode immediately. Then, I apply Strip Silence on both clean tracks. This will take any section of audio below a given loudness threshold and replace it with a blank hole. This lets me quickly go identify and drag over leftover bits of noise that didn't get picked up by the noise reduction in the audio processing stage to delete them.

I usually listen to the episode from start to finish, paying close attention to the edges of those voice clips, removing any umms, ahs, and mouth noises that sit on those edges. Over time, you will develop familiarity with the waveforms of these things and become able to do it by eye instead of needing to listen to the entire audio, speeding up the process significantly.

In the case of crosstalk, triple-tapping on an audio clip will select that clip and anything to the right of it on all tracks, letting you easily shift overlapping audio to make things sound more natural. This will naturally desync your audio from the dirty tracks sitting in your editor, but your clean tracks will remain synced with themselves as long as everything was properly lined up at the start of editing.

When the episode is done, I delete the dirty tracks entirely and select all of the remaining audio clips. I then use the Tighten command to give silences a maximum length so things feel more conversational, and so that any bits I chopped off from the starts and ends of voice clips don't lead to longer awkward pauses.

Once Tighten is done, I drag the audio all the way to the left of the timeline so the leftmost clip begins at 0:00. Now we can move on to metadata and exporting.

Exporting

This step of the process varies depending on whether or not the episode in question calls for the use of chapter markers.

If the episode is being released without chapters (most weeks), I will tag the metadata for the episode directly in Ferrite, and use the Export functionality to generate a mono 64kbps CBR MP3, which I usually send up to Dropbox for Luc-Olivier to publish later in the weekend.

If the episode is being released with chapters, I won't tag any metadata within Ferrite, and I will instead export a WAV file of the entire episode and send it over to my Mac via AirDrop or iCloud Drive. There I use Forecast to tag the chapter metadata and export a mono 64kbps CBR MP3, which I will then put in Dropbox for Luc-Olivier to publish.