This just in by way of ReadWriteWeb, Google has recently announced the release of a new Labs tool – known as GAUDI, or Google Audio Indexing – that enables you to perform text searches of the audio content featured in video clips.
Think about that for a moment. This is not to be confused with searching the text-based descriptions of the videos, but the audio tracks themselves.
As the Official Google Blog explains (“In their own words”: political videos meet Google speech-to-text technology):
“With the help of our speech recognition technologies, videos from YouTube’s Politicians channels are automatically transcribed from speech to text and indexed. Using the [iGoogle] gadget you can search not only the titles and descriptions of the videos, but also their spoken content. Additionally, since speech recognition tells us exactly when words are spoken in the video, you can jump right to the most relevant parts of the videos you find.”
The reality of this news is significant.
Traditionally locating content on the web has required the use of a search engine, but in order for the search engine to recognise a webpage, as well as the content within it, it had to be indexed. There are several different ways to do this, but by and large this relied on the presence of text. When someone was attempting to locate information on the term “sheep” the search engine would cross reference its massive indexes for all content that had the word “sheep” in it. When it found something the page or file would be included in the search results.
This was all fine and good when it came to static text – which was the primary constituent of the web in the early days – but along with the increasing usage of images, audio and video, things began to get more complicated. There are ways to provide textual descriptions of these sorts of things in the file’s metadata, by by and large the information provided is pretty basic stuff and hardly comprehensive.
So in reality, unless each audio or video file was accompanied by a transcript you wouldn’t have a clue what was actually in it, let alone a way to search for it.
The significance of Google’s announcement is therefore hard to overstate. While the scope of GAUDI (Google Audio Indexing) indexed material is currently small and limited to political video clips, is It also a proof of concept for what we can hopefully expect in the future. Consider, for example, the opportunity to search the entire audio contents of the entire YouTube catalogue. Viewers would no longer be limited to the frequently useless descriptions that accompany each video; they could instead search the audio contents of the files itself.
Furthermore there’s no reason why this technology couldn’t be applied elsewhere, to non-Google affiliated sites. Google has made it its mission to make all the world’s information searchable after all.
In an educational sense, the increasing popularity of podcast lecture recordings would stand to benefit immensely here, since students would conceivably be able to locate specific topics across the catalogue of their course’s digitised materials AND be able to click directly to the portion of the file that discussed the term or phrase.
While it may be true that similar technologies already exist, there is little doubt that a solid offering by Google would bring the idea into every day use. It will be interesting to see where this goes.
In the meantime if you’d like to experiment with Google Audio Indexing, you can do it here: http://labs.google.com/gaudi
