Podcasts Have A Serious Reach Problem with Zac Cichy

Zac Cichy returns to round out the first season of PTS. We discuss a number of topics: quitting Red Bull after some serious health issues, the long-awaited launch of Whole & Part, podcast transcription and theories about the future of wearables. 

A transcript of part of the discussion about podcast transcription. How meta:

[50:55]
Ben: They're competing with apps. They're competing for three minutes of someones time and you're putting something out there thats an hour.

Zac: Yeah.

Ben: To get them in the door. To me thats whats interesting and what the big problem seems to be - getting your foot in the door so someone will give you any, any of their time at all before you ask for hours of it. And its, its tough.

But what you're doing that I really like are these transcriptions.

Zac: Yeah, the idea sort of came about because I had this crazy idea of creating full blown transcriptions of The Menu Bar every week. And we may do that, we may not. Its a lot of work. Any time you do transcriptions you have to figure it'll take 3, 4, 5 times the amount of time as it takes you to listen to the episode. And a lot of shows we they tend to be over an an hour long.

You're talking about sitting down for 4 to 5 hours just to do a transcription. At least to do it well. You could probably blow through it a lot quicker but there would be a lot of errors.

Ben: And its hard to automate.

Zac: Its nearly impossible from what I can tell. One trick that you could do, is if its a podcast, say its me and Andrew. If we have Dragon Dictate on our computers and Dragon has gotten to know our voices. Then, then you can kind of automate it. But the problem is you can't just take random audio from just anybody where something like Dragon hasn't gotten to know you.

You can't just take that throw it in and expect to get good results. 

Ben: I tried that. I played around with that, with automating some slicing up in Logic.

Zac: Yes.

Ben: I didn't have Dragon, so I tried just using Dictation and it was awful and I figured out that would be the problem. And theres another problem that you still need to check it. Thats still where it takes time anyway. And you're still going to have the ums and ahs. 

Zac: Right. So I mean there's been a few times where you and I've kinda  talked over each other a little bit, and you have to make a decision if you're doing a transcription whether you're going to leave it in or leave it out and splice it up in some other way. And thats a thing that with a computer, no matter what, we're not there yet. We don't have AI that can parse that stuff.

Yeah, it becomes pretty much necessary to just do it the old fashioned way. Go through and play, rewind, play rewind, play rewind.

Ben: But thats just it, thinking about the limitations and the focus. Thats whats interesting to me, I think I mentioned this in the post I wrote the other day is I think it what you're doing may be better than putting the whole thing up. I would rather you take the time to think about whats really good and you can present a thing, or three things that can be a lead in.

Zac: Sure, um, yeah, the problem is in an ideal world, where is everything as I want it to be - what would be great is to have an app that had a full transcription of every single episode and a perfect timestamp of everything. Kindle audiobooks actually do this now, it tracks along with speech. 

And it works very well, in fact I need to look into that its very clever.

But yeah I mean in an ideal world thats how podcasts would be. Because... podcasts have a serious reach problem. Nobody has any good data on this. Nobody knows how much podcasts have grown over the past few years. I suspect that its grown but nobody knows really how much. 

Who's really getting into them? It seems to me its still geeks and nerds for the most part that still get into it. Maybe we're attracting even more geeks and nerds. But as far as everybody else podcasts still aren't even a blip on most peoples radar. 

So yeah, transcription is definitely one thing that I think could help with that. So thats what I've been doing. Any time I'm listening to a show and theres one bit that I think is particularly good, like with The Prompt there was a part where Federico rants a little bit about Google and I thought that was just a really good chunk and I decided "OK, I'm just gonna go through and timecode this and transcribe it."

Its a different kind of way to do a like a link blog I think. Its a different kind of content I don't think people have really seen before. And yeah it provide value to - I have a lot of followers on Twitter who say they just don't listen to podcasts. Maybe they don't have time, but  they do feel like they're missing out on things.

And really its no one takes the time to sit there and transcribe this stuff. And really, with good reason, its takes time. Its not easy.