A Plea for Transcripts

I enjoy listening to podcasts about design and I’ve posted quite a few to this blog but I’ve noticed a disappointing gap in podcasts as a service. They never seem to provide transcripts.

Podcasts are engaging, but the human voice is a terribly low-bandwidth channel for communicating information. It’s much too linear. Podcasts are difficult to skim or summarize and impossible to search. Transcripts are invaluable from an archival point of view. They’re de rigeur for news and political websites but I didn’t realize what I was missing until I stumbled across the transcripts for the Intersections 2007 podcasts last November. For service designers I think this should set the standard.

I’ve been experimenting with ways to make podcasts more useful for my own needs. At the very least I try to summarize key points and match them with timecodes. I’ve also started transcribing the podcasts I listen to before they make their way to my blog. But I shouldn’t need to do that myself. Transcription should be a service that goes along with recording the interview in the first place, like editing or encoding an MP3.

Transcriptions take some time to do well (at least three times the audio length) but they’re not terribly difficult. My first transcript was for the Dick Buchanan keynote a few weeks ago. After about five minutes of clumsily switching back and forth between the playback controls and the keyboard I decided to build a tool to help speed things along. It’s a transcription pedal made from a Wii controller that allows me to keep my fingers on the keyboard while I control playback with my foot.

It’s pretty easy to cobble one of these together (it’s just foamcore and some tape). I sync the Wii controller to my MacBook and control iTunes with a tap of the A button using a system called WiiMotion to make it all work. It’s overkill for this kind of simple application (and it requires some familiarlity with the terminal) so if you’ve got an Apple Remote it might do in a pinch.

My point is, transcripts should be the standard. It’s not just an accessibilty issue and it’s not about altruism (though it’s pretty good karma). If podcasts are easier to consume then more people will consume them. Transcripts help drive traffic by exposing content to search engines. Bloggers can quote the text. Readers can skim to decide whether a podcast is worth listening to. Ultimately it’s about increasing consumption, and that’s the point of posting to the web in the first place.


  1. In one of his podcasts, Jared Spool mentioned that they use Mechanical Turk for transcribing podcasts.

    I’m sure it’s cheaper than doing it yourself (assuming you own time is pretty valuable).

    http://www.mturk.com/mturk/welcome

  2. Jeff

    That’s an interesting idea. I did some searching and came up with a transcription company called CastingWords using the Mechanical Turk.
    http://castingwords.com/store/

    Here are a few with more traditional models:
    http://enablr.com/transcribr.php
    http://www.escriptionist.com/transcription-rates/
    http://www.connecticutsecretary.com/transcription.htm

    Looks like they’re around $1 to $1.50/minute for simple transcriptions with one week turnaround. That’s less than $100 for a one-hour interview. Twenty bucks for some of the shorter podcasts.




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s