Podcasting is a funny animal.  As much as we’ve heard about how audio (and video for that matter) will dominate content on the net, it isn’t happening.  Text is still king, even though talk podcasts consist of that same text just in speech format.  So why aren’t we seeing podcasting making it to mainstream?  There’s one commanding reason:  searchability.

Finding information inside a podcast is, for the most part, impossible.  We do have ways of compensating, typically by adding a human element with show notes or tags.  For some this is sufficient in that, if done perfectly, the show notes should hit every key word in the podcast.  The tags would then be picked up by Google and the other search engines leading listeners to that podcast.

But there are two major flaws with this ideology.

• You cannot tag every possible key word. If you did, you wouldn’t be tagging, you’d be transcribing.  As a result it is inevitible that somebody will search for a key word in your podcast that they will not find.
• The information is too difficult to get to. Unless looking for a podcast in particular, a typical Googler will not download a 20mb – 60mb file to find the information they were looking for.  Likewise, they’re not going to listen to 20 – 60 minutes of content to find the information.  They may have tagged the podcast appropriately, but the searcher would still have to download the entire podcast and find the part where they talked about the subject they were looking for.

Audio information simply requires too large of a time investment to reach the masses.

Audio information simply requires too large of a time investment to reach the masses.  The answer that’s been floated around for quite some time is transcription.  But is it really an answer?  Some say no, and their arguments are valid.  Consider the benefits:

• Transcription takes time.  If you have a 60 minute podcast, it will take you at least 60 minutes to transcribe it, and that’s considering you can type at the speed of speech.

and/or

• Transcription costs money. There are transciption services out there such as Casting Words that will charge anywhere between $0.75 and $2.50 per minute, depending on how quickly you need it.  Consider that 75 cents per minute still costs $45 for a 60 minute episode and could still take over two weeks for it to be completed!  If you’re willing to manage the transcription directly, you can find plenty of willing transcriptors at Amazon’s mturk which could save you considerable amounts of money.

How do you provide the user exactly what they’re looking for without forcing them to download the entire podcast?

However, this argument makes one large assumption.  It assumes that the transcription will only be completed if you initiate the action.  With the progress that’s been made in online collaboration and crowdsourcing, podcasters have at their disposal an army of volunteers eager to help in any way they can, often short of giving money.  Several podcasts have toyed with the idea of setting up a wiki to have their listeners transcribe the podcast for them.  While promising and likely the preferred method of podcast transcription in the future, it still doesn’t solve the entire problem.  How do you provide the user exactly what they’re looking for without forcing them to download the entire podcast?  How do you aggregate podcast transcripts into a coherent repository from which to search?  How do you protect the transcripts to keep them from being used in blogs across the internet?

And that’s where I’ll leave this post.  The question whether or not to transcribe a podcast is an open one.  I agree with most when they say they don’t see the benefit of transcription compared to the effort and cost it requires.  There are several new projects we’re working on to solve these problems, but until we can decide whether or not they are even viable, the problems remain.