Multimedia search firm Blinkx is the latest to use Really Simple Syndication (RSS) to make it easier for people to find the web video and audio they want.

Users can look for any word uttered in news videos, podcasts and video blogs.

Multimedia search is beginning increasingly important with Google and Yahoo also providing video search.

Text analysis

Blinkx's service uses smart voice recognition technology to scan video and audio it finds on the web. A searchable transcript is then automatically generated.

With RSS built into Blinkx's service, Harry Potter fans can save a Potter search and enter it into the webfeed reader.

Blinkx founder, Suranga Chandratillake, told the BBC News website said that because there was so much audio and video on the net, particularly with the advent of podcasting, it made sense to make search and delivery automated.

"What we did there was look at where search today fails," he said.

"One big area is TV and video content. There is lots of it but it is surprisingly difficult to find it."

You can also check : Top News News of the Week

Although video and audio on the web has been around for some time, the rise of broadband and better compression techniques have driven its popularity and quality.

Most main news websites carry video reports now. Blinkx, a year old this week, has struck deals with the likes of the BBC, Bloomberg, CNN and Fox to index the content.

There is also a small but growing band of video bloggers now who are also threatening to take podcast-type programmes into the visual arena.

Poor descriptions

The video search services offered by Google and Yahoo do not analyse the content with voice recognition or play the programmes. They search description information from host sites, or closed caption information.

US broadcasters and the likes of the BBC are obliged to include closed caption information in broadcasts as part of a commitment to accessibility.

But, explained Mr Chandratillake, it is often missing in older, archival content.

Often publishers do not provide adequate enough descriptions of the what is in web video either, he said.

Blinkx uses voice recognition technology too, which has been around for some time but is becoming smarter and more accurate.

The speaker-independent voice recognition technology, from UK-based Autonomy, is 96-97% accurate on broadcast quality video, according to Mr Chandratillake.

The same "listening" technology is used by the Department of Homeland Security in the US as well as other security agencies.

"Our core back end server is an IBM machine which cost around $1,500. One of them can listen to four 'conversations' in parallel," he said.

"One of the content types we have added recently are podcasts. That alone is generating 500 hours a day of content to search."

My media

Blinkx monitors nearly 27,000 podcast "channels" or feeds, a jump from just over 1,200 in January.

There are other services popping up, such as Podscope, which search the content of podcasts, rather than services which list them in directories according to genre or popularity.

What Blinkx and others are doing with automated search and delivery is yet another step closer to people defining their own media channels.

"It is allowing media content to be atomised and split into smaller chunks to be passed around to different devices," he explained.

"People will expect more and have more control over when, where and what they watch."