An impor­tant but often-overlooked part of cre­at­ing an audio file for pod­cast­ing is prepar­ing the transcript.

Hav­ing a tran­script is impor­tant for acces­si­bil­ity rea­sons, but also because if they’re any­thing like me, your audi­ence may not have time to lis­ten to the full audio and or might pre­fer to skim through the tran­script to see what your pod­cast con­tains. There’s been some debate recently on whether the num­ber of pod­cast down­loads reflects the num­ber of peo­ple who actu­ally lis­ten to the audio itself. I have a direc­tory on my com­puter chock-full mp3s which have been par­tially or com­pletely ignored which would lend weight to the argu­ment that many peo­ple think pod­casts are a good idea, but just don’t have the time to lis­ten to them.

In the case of a record­ing of a live pre­sen­ta­tion, prepar­ing a tran­script — writ­ing down every last thing that the speaker said — sounds like a lot of effort, espe­cially if you weren’t the speaker. How­ever, for US $0.42 per minute, you can get some­one else to do the hard work for you at Cast­ing Words. I did this for the March Web Stan­dards Group meet­ing in Perth, partly as an exper­i­ment to see how it turned out, and I’m really impressed with the results.

Order­ing the transcript

After prepar­ing the audio files, I uploaded them to my site and filled out an order on the Cast­ing Words web site. They take either indi­vid­ual audio file addresses, or an RSS feed for larger orders. They sup­ply you with an RSS feed address which you can use to track the progress of your order, but as mine was small (2 files) it’s less use­ful than it would be for, say a whole con­fer­ence worth of audio casts.

I made my order Fri­day night my time, and was impressed to find the first tran­script in my email when I checked Sat­ur­day morn­ing. The sec­ond email arrived around 18 hours later, on Sun­day morning.

Tran­script format

The tran­scripts arrived as text in the email, with links to rtf, plaint text and html ver­sions on their web site. The HTML was marked up with cite and block­quote ele­ments, which was nice, but with some incor­rect nest­ing so it did not val­i­date. Hav­ing a HTML ver­sion as a bit of a bonus as I was expect­ing to have to mark it up myself, but I did have to do quite a bit or rework­ing to fix the nest­ing and remove some classes that were also in there.

Qual­ity of the transcript

It was never going to be an easy task: us Aussie tend to speak a bit funny, after all. But on the whole, the qual­ity of the tran­scribed text was excel­lent. With the tech­ni­cal top­ics, I fully expected to have to make some cor­rec­tions — I actu­ally conned the speak­ers into doing it them­selves — but sur­pris­ingly lit­tle had to be done. We did a bit of surgery on Nick’s to take out some parts which weren’t rel­e­vant to the topic — there were some issues with the mouse he was using, and the pre­sen­ta­tion kept jump­ing back to the start. I think some edit­ing is accept­able as the pur­pose of this kind of tran­script is to pro­vide an alter­nate media ver­sion of the audio, not a word-for-word record of what was said.

To sum­marise…

Over­all the Cast­ing Words expe­ri­ence was excel­lent and I would rec­om­mend the ser­vice to any­one who wishes to pro­vide tran­scrip­tions of pod­casts. It was very afford­able — the whole exer­cise cost me AUD $17, for a lit­tle over 30 min­utes of audio. The results were speedy and accu­rate, less than 36 hours all up with very few alter­ations required — and while it would have been nice to cut and paste the html markup directly into my page with­out hav­ing to edit it, hav­ing mul­ti­ple ver­sions of the out­put was handy.