Yes, Colin, seems expensive. And perhaps you hit upon a process (Google's captioning) + 1/2 hour work gets you there. And since your time is only worth around $10US/hour, you have the optimal cost solution. 😀
When I did an investigation a few years ago, for me to achieve Rev.com accuracy for a 20-minute video took me around 3 hours. And I earn big-big bucks!!! 😇 So the Rev.com cost was the most cost-effective solution.
Incidentally, the cost of having outsiders doing captioning has been investigated a lot by my university. So far, Rev.com is still the chosen solution. But this can change quickly if some other company could outdo Rev.com. I am looking forward to the day that smartphones do this automatically for any spoken word, then we will not need to administratively create captions. I do believe that this will happen, just as smartphones (tablets, and computers) can now "read" words.
Also, as a point of reference, Rev.com does provide "auto-captioning" for $.25US/minute (instead of $1.25US/minute), but I tried it and found it to be too inaccurate.
(I am not associated with Rev.com, nor do I own any financial interest in Rev.com. No "ad" intended in my post.)