Frequently asked Questions

Informs you of typical asked questions

WhatsNew:

Innovation

March 14, 2013

Innovation is the key to our recent product releases.

demo

CurrentNews:

New Releases

March 14th, 2013

We welcome you to our new exciting web site. The new site utilises the latest technologies in dynamic web site design. It demostratates the ultimate in portal utilisation. This site allows it's visitors to enjoy a tour of the RODI world of media technologeis.

demoThe site is hosted by the latest Azure Cloud server and alllows high speed interaction when examining our newest and innovative solutions.

RecentEvents:

Innovation

March 14, 2013

The company has released a suite of new products that will revolutionize the way organistaions operate. Support for Law Enforcement and Media Video Mining are the key areas where the new products are focussed.

 

demo

 

Frequently asked Questions

 

1. What is a Large Vocabulary Continuous Speech Recognition (LVCSR) system.?

LVCSR search uses a completely different approach over traditional phoenetic systems. It uses a speaker independant speech-to-text technology making use of specilized vocabularies or topics.

The solution provides full vocabulary support for Medical, Law Enforcement, Legal, Broadcasting and Government environments.

Enable Text Search of Audio / Video Files. The ability to use text keywords and phrases to automatically search Audio /Video files is now within reach for organizations of any size. Audio / Video Mining eliminates the time and cost associated with manually indexing rich media, enables the indexing of 100% of the speech information within Audio /Video files. By using an advanced speaker-independent dictation engine, Audio / Video Mining creates an XML speech index data for every word spoken within an Audio /Video file. The index data includes word, time stamp, confidence levels and metadata associated with the speech information, and can be created from any quality broadcast and telephony-quality sources.

  • Support for Audio Streaming. Recognition can now be performed from a streaming file, eliminating the need to produce an Audio / Video file.
  • Telephony and Broadcast Acoustic Models work in conjunction with complex language analysis to produce unsurpassed results. The models are speaker-independent, meaning they’re built to support recognition for an unlimited number of speakers with different voices and accents.
  • Recognition Confidence Score. Accuracy levels depend upon the quality of the recording. Studio-based content will provide higher accuracy levels, but the system also provides a reasonable level of accuracy for telephone, public presentation and broadcast content.
  • Increased Accuracy. The system recognizes "all words," not just keywords. The accuracy of preconfigured vocabularies can be further fine-tuned using the Vocabulary Tool to include organization-specific terms and proper names. This tool automatically customizes vocabularies with unique terms, such as industry-specific terminology or topics, resulting in outstanding recognition.
  • Enterprise database support allows indexing of unlimited number of media files. Typical retrieval times "hits" are less than 1.5 seconds.
  • Full streaming of Audio / Video files reduces network bandwidth demands.