Watching Your Words

By Ian, on September 25th, 2006

Techcrunch’s Marshall Kirkpatrick reveals an interesting new technology development designed to improve the podcast format:

Seattle based podcast discovery and management service Pluggd is unveiling a major new feature at DEMO this weekend that combines speech recognition and semantic analysis to let users search for and skip to parts of an audio file that are related to topics of interest to them. Itâ€™s more than just speech recognition.
This is one of the most compelling examples Iâ€™ve seen lately of a growing trend: making multimedia content more granular and letting users take even greater control over the media we consume. We donâ€™t just want to consume what we wish, we want to consume it in the way we wish.

Podcasts have a number of ease-of-use problems that this helps with. Only Apple’s podcast format can be properly split into chapters, for example, but those are less popular because you can’t play them on non-Apple devices. On an iPod, you can’t fast-forward an MP3 file. Lastly, podcasts can only be consumed in real-time — you can’t skim through it the way you can with printed publications.

However, while ease of use is highlighted, I think the impetus for search technologies in audio comes from other motivations too. For publishers, podcasts (and video) have proven difficult to monetise, with most opting for a flat sponsorship deal, if anything. However, if you know what words are in a piece of audio, or video, then automatically targeted advertising works a lot better. You could opt, for example, for your adverts to be appended to podcasts that contain the word ‘iMovie’, the same way that advertisers buy these words from AdSense. Advertisers might also be able to generate a blacklist of offensive terms and competitors’ names to avoid being associated with content they would rather not be.

There has been talk of metatagging innovations in audio and video, but the problem with all metatags is that they can be spammed. True audio search is superior in every way. The video equivalent is already underway at Dave.TV which will apparently monitor the words in a clip to decide on appropriate ads:

We will be using the same technology used by Homeland Security to monitor [telephone] chatter. Audio keywording will allow us to contextually figure out where to sell ads and to place more than just pre– and post-roll ads.