Podcast workflow

From nothing ➡️ published podcast episode:

  1. Have the call
    • Use Riverside.fm. It locally records each person in high quality and uploads them as it can
  2. Load the call into Descript
    • Download the files from Riverside.fm
    • Make a new project in Descript
    • Add all the downloaded files, as a multitrack sequence, and name each speaker. It’ll take a few minutes to transcribe
  3. Clean up the audio
    • Add the ‘Studio Sound’ audio effect to improve audio quality
    • Click ‘Detect Transcription Errors’ to automatically improve the transcription
    • Edit out anything confidential
    • If filler words / gaps / anything else needs reordering or removing, do that
  4. Clean up the video
    • Set the template to the one with 3 floating bubbles. Right click the biggest bubble -> replace with, so that it contains the guest
    • Right click the scene thumbnail and click Copy
    • Use ‘Add scenes by active speaker’ to have it switch to the active speaker throughout
    • For parts where you want everyone on screen (intro, outro, and maybe scenes where everyone’s speaking over each other / laughing together) right click -> Paste the 3-person template you copied
  5. Publish
    • Audio: click Publish, then Export tab, Audio. Save the file
    • Transcript: click Publish, then Export tab, Transcript. .txt file. Exclude everything except Speaker labels and Speaker label timecodes. Save the file
    • Upload the audio & transcript to the Transistor.fm dashboard for syndication to all the podcast players. Add title and show notes
    • Video: click Publish, then Publish tab, YouTube. Paste the title and show notes