MARTHA

The broadcast media industry is undergoing rapid change. 24x7 news is now commonplace, requiring newsrooms instantly to combine feeds from all around the world with their large scale media archives.
Emerging technologies in phonetic audio search can find matches to any words and phrases (including names) in spoken audio, creating the potential for journalists to be alerted about breaking news and to rapidly search their archives for relevant content - without being limited to just catalogue metadata.

Project MARTHA brought together from the UK a national news broadcaster, a leading media asset management (MAM) company and a pioneering phonetic audio search company to solve the challenges needed to create a prototype journalist alerting and large scale archive searching system.

Innovative features of the prototype that has been developed include:
- Searching feeds using spoken audio as the basis of the search, which is very novel in this industry
- Using phonetic audio search - which can find names of individuals, places and companies
- Real time journalist alerting, which is totally new in the broadcast industry. MARTHA provided the first demonstrated use of audio alerting technology to find news stories across multiple feeds as they happen
- Application of large-scale audio search to media archives, which involved developing novel technical architectures for rapid search.

The prototype alerting and archive search system was installed and evaluated by journalists. Both the archive-search and the real-time alerting functions were very well received, with all feedback being positive. The system was able to find audio and video clips that would otherwise have taken much longer.

Spin-off benefits from MARTHA include the potential to apply the technology beyond news, to widen intelligent access by consumers to context relevant content, including internet-based media and democratised content.