There’s a lot of talk about artificial intelligence dipping its robotic toes into the pool of humanity lately, what with a certain evil tech overlord and his self-driving cars terrorizing the streets of Toronto.

But artificial intelligence isn’t just wreaking havoc in the streets… it’s also wreaking havoc… in our podcast production processes.

If you work in podcast production, there’s a good chance you’ve tried Descript once or twice. For those of you who haven’t had the pleasure, Descript is a transcription platform that will cut your audio based on how you manipulate the transcript.

I feel like it doesn’t make sense without showing you what I mean, so here’s an example. You feed your audio into Descript, and it creates a transcript for you.

It’ll look something like this:
 

Speaker One: You know, I’ve been reading Vocal Fridays a lot recently, and I have to say, it’s the best part of my week.

Speaker Two: I thought I would miss Michal’s writing when she was in Europe, but I actually really liked reading newsletters by the other Vocal Fry producers. 

Speaker One: Yes! They’re beautiful writers and have really good insights about what’s going on in the world of podcasting. 

So nice of our guests to say that. Stop it! No you stop it. 

Once we have our transcript, then we can manipulate it. If I wanted to cut this into making our guests say what I want them to, I’d strike through the words I want to eliminate, like this: 

Speaker One:  ̶Y̶o̶u̶ ̶k̶n̶o̶w̶,̶ ̶I̶’̶v̶e̶ ̶b̶e̶e̶n̶ ̶r̶e̶a̶d̶i̶n̶g̶ ̶V̶o̶c̶a̶l̶ ̶F̶r̶i̶d̶a̶y̶s̶ ̶a̶ ̶l̶o̶t̶ ̶r̶e̶c̶e̶n̶t̶l̶y̶,̶ ̶a̶n̶d̶ ̶I̶ ̶h̶a̶v̶e̶ ̶t̶o̶ ̶s̶a̶y̶,̶ ̶i̶t̶’̶s̶ ̶t̶h̶e̶ ̶b̶e̶s̶t̶ ̶p̶a̶r̶t̶ ̶o̶f̶ ̶m̶y̶ ̶w̶e̶e̶k̶.̶

Speaker Two:  ̶I̶ ̶t̶h̶o̶u̶g̶h̶t̶ ̶I̶ ̶w̶o̶u̶l̶d̶ ̶m̶i̶s̶s̶ ̶ Michal’s  ̶w̶r̶i̶t̶i̶n̶g̶ ̶w̶h̶e̶n̶ ̶s̶h̶e̶ ̶w̶a̶s̶ ̶i̶n̶ ̶E̶u̶r̶o̶p̶e̶,̶ ̶b̶u̶t̶ ̶I̶ ̶a̶c̶t̶u̶a̶l̶l̶y̶ ̶r̶e̶a̶l̶l̶y̶ ̶l̶i̶k̶e̶d̶ ̶r̶e̶a̶d̶i̶n̶g̶ ̶n̶e̶w̶s̶l̶e̶t̶t̶e̶r̶s̶ ̶b̶y̶ ̶t̶h̶e̶ ̶o̶t̶h̶e̶r̶ ̶V̶o̶c̶a̶l̶ ̶F̶r̶y̶ ̶p̶r̶o̶d̶u̶c̶e̶r̶s̶.̶ ̶

Speaker One:  ̶Y̶e̶s̶!̶ ̶T̶h̶e̶y̶’̶r̶e̶  beautiful  ̶w̶r̶i̶t̶e̶r̶s̶ ̶a̶n̶d̶ ̶h̶a̶v̶e̶ ̶r̶e̶a̶l̶l̶y̶ ̶g̶o̶o̶d̶ ̶i̶n̶s̶i̶g̶h̶t̶s̶ ̶a̶b̶o̶u̶t̶ ̶w̶h̶a̶t̶’̶s̶ ̶g̶o̶i̶n̶g̶ ̶o̶n̶ ̶i̶n̶ ̶t̶h̶e̶ ̶w̶o̶r̶l̶d̶ ̶o̶f̶ ̶p̶o̶d̶c̶a̶s̶t̶i̶n̶g̶.̶

When it comes time for me to export the tape, I would get an Audition session that’s just the phrase:

“Michal’s beautiful.”

You guys! I’m blushing!

In theory, Descript should simplify the first pass of audio editing. It basically allows you to combine a paper edit — when you go through the transcript and figure out what you want to cut — and the first version of actually cutting the tape.

Except… Descript keeps adding updates.

And now it seems like they’re actually more interested in being a tool for video content creators… and maybe they never cared about podcast producers to begin with.

The truth is, I have always been a Descript skeptic. I have always found it clunky, and their cuts always need fixing. They cut out a breath where they shouldn’t. The edits sound too obvious. Sometimes I wonder if it takes me twice as long to fix their errors than it would to just cut the tape myself the first time. It’s just missing something… human.

Audio editing has changed a lot. We talk about “cutting tape,” but we’re working on DAWs (digital audio workstations) never once needing to pull out a razor (or a knife? I don’t know what I’m talking about) and splice a physical reel. When I talk to my friend’s mom about when she started working in radio, she talks about being in a room and, yes, literally cutting tape and sticking it together. That must have taken so long and been so finicky. How did anyone ever get anything done?

But while digital tools help us get more work done, faster, at the end of the day, humans are still the ones taking hours of tape and making it into something wonderful. AI just isn't good enough to do what we do… yet.

We have a new writer on the blog this week, Jess Schmidt, who used to work as a closed captioner for live television. And Jess is here to make the argument that AI software can't beat human transcribers. We love a rant here on Vocal Fridays, and I encourage you all to read Jess’s.

Here's what else we have our eyes on this week: 

The New York Times made this extremely cool guide to creating spatial audio. Audio nerds, get thee to the Times!

Have you subscribed to Kattie Laur’s newsletter, Pod the North, yet? She just told me about some very exciting stuff coming down the pipe and readers are in for a treat. In case you missed it, her most recent newsletter featured an interview with Falen Johnson.

Holiday fun! On the internet! The Sonar Network will be celebrating its five-year anniversary next Sunday, December 11 with a TWELVE (12) hour comedy livestream! It’ll be streaming on their YouTube channel from 12 pm to 12 midnight, and you can check out the lineup on their Instagram.  

CBC Radio is hiring a temporary, full-time associate producer for Metro Morning, the flagship Toronto morning show. Apply by EOD December 9.

CBC Halifax is hiring a permanent, full-time video editor. Apply by EOD December 8.

CBC Calgary is hiring a permanent, full-time producer to work with network talk radio.

The Globe and Mail is hiring a senior producer for The Decibel, the Globe’s daily news podcast. You’d be leading a wonderful team of four audio journalists and a host – dream job for people who love daily news!

Pacific Content is hiring for two roles: a part-time producer and an audience development specialist. One person to help make the podcasts, and another to help get them into people’s ears. Two paths to go down… which one will you choose?

JAR Audio is hiring a fully remote podcast marketing coordinator. The salary is listed as $50k.

The BBC is “aggressively expanding” in North America. Get in touch with Saeed Ahmed, head of digital journalism.

Paging women and non-binary folks in audio drama – Parkdale Haunt wants to connect! Send them a dm, make some friends!

We’re all just out here fighting the good fight.

We’re reading Jess Schmidt’s blog post about why AI can’t beat human transcribers, and breathing a sigh of relief that we’re not redundant… yet. 

We want to hear from you! What are you looking for in your podcast news? Let us know on Twitter, Instagram, or by email at info@vocalfrystudios.com.

Thanks to Emily Latimer for editing this newsletter, and to Katie Jensen for designing it.

We’ll see you again on December 9. Until then, here’s an update from producer Kattie Laur’s dog, Joe, looking like a groovy jazz musician in his wayfarers.

Yours in friends and fries,
Michal