Open Source Podcast Mastering with Audacity

Published July 2, 2023

I haven’t mentioned it much, but I’ve had a casual podcast with a friend since late 2018. We don’t upload often, but it’s been a fun low-stakes side project. If you like reading this sort of stuff, you’ll probably like Avocodo.

Anyways, from the beginning we were doing it as scrappily as possible:

1. free platform hosting via Anchor (now Spotify for Podcasters)
2. free¹ file storage with a .edu Google Drive
3. free recording & editing with Audacity

Figuring out how to leverage Audacity for our needs was certainly a learning curve where we pieced together the workflow and certainly failed a few times as we improved. Podcasts seem to still be the hip thing, so I’m hoping by doing a semi-polished copypasta knowledge dump that I’ll be paying it forward.

One final caveat before the dump: you’ll still want some decent recording hardware like a discrete microphone². There’s a good chance you’ll also benefit from an inexpensive USB audio card.

Now, without further ado.

Recording

• Mono, 44100 Hz
• Mic check (input sensitivity, physical position)
  • Record a short sample clip and see if you clip during loud bits of conversation
  • If you have a (mechanical) keyboard, is it obnoxious in the background?
  • Do the sounds of your breathing come through?
• Export: MP3, constant @ 192 kbps, force export to mono

Mastering

1. Align tracks based on audio cue (when the levels spike)
2. Check late in recording for clock drift
   1. Find an audio source late in recording to reference & either do math or guess (typically << 1% drift)
   2. Fix with Effect > Change Speed (select single track w/ double click)
3. Noise removal if necessary w/ Effect > Noise Reduction
   • Good ref: https://theaudacitytopodcast.com/tap012-how-to-remove-noise-with-audacity/
4. Using Chris’s Dynamic Compressor select both tracks w/ Ctrl/Cmd + a
   • Good ref: https://nototally.com/audacity-tutorial-compression-using-chriss-dynamic-compressor-plugin/
   • Approximate levels, will need tweaks per recording:
     • Ratio: .9
     • Hardness: .55
     • Floor: -18
     • Falloff: 2
     • Amp: .9
     • Model: 0
5. Final touches:
   • Can add silence (Generate > Silence)
   • Any bleeps & stuff (Generate > Tone [sine, 1000, .8])
   1. Trim beginning & end
   2. Add short Fade In/Out to beginning/end (Effect > Fade In/Out)

Publishing

This certainly less useful for the general population but I’m including it in the spirit of openness and keeping parity with the source document.

• Title: #No (0-indexed): Title Case Title
  • Ex: #6: Return from Hiatus: ASMR & Crypto
• Copy in show notes from episode doc
• Advanced options:
  • Season 1 for now
  • Episode 1-indexed (e.g. #6 becomes episode 7)
  • Full type
  • Explicit always (in case of swearing)

NB

My final note would be that this is not a time-friendly process. I typically find that I spend twice the length of the raw recording time in the editing and publishing stage of the pipeline. For example, if I spend 2 hours recording raw audio, then I typically spend another 4 before it’s fully published.

This is partially due to my editing style: I end up listening to the recording in full as I make edits end-to-end. There are certainly optimizations to be made. But I also personally find it a productive use of my time and enjoy it overall.

Last modified July 1, 2023  #documentation   #words