Audio Analytics

Brochure

Medirva™ – Best in Class Audio Analytics with Workflow Management

Medirva™ provides best in class data mining tools to allow rapid identification of important information that may be buried within thousands of hours of recordings together with workflow management to allow you to optimize the use of automated and manual resources. Whether you are undertaking transcription, searching for a word, a phrase, a person or just trying to understand complex relationships Medirva™ voice recognition can greatly enhance your operational capability and save significant costs.

With ever increasing communications and the decreasing costs of mass storage huge quantities of data from a variety of sources are being recorded. The result is that most organizations do not have sufficient manual resource to analyze the recorded data and may be missing vital information and intelligence. Data mining tools can be a very effective solution to these problems, but only if certain conditions are met. Medirva’s™ patented resource allocation technology allows Supervisors and Managers to determine when to use which audio analytic techniques to achieve the best results.

With the facility to select from key word spotting, speech to text, speaker ID, language ID, auto gisting and visual presentation tools, you can use the most effective tools to rapidly get vital information at critical times.

Medirva’s™ easy to use but secure web browser interface means access to data anywhere, anytime with no more requirements for complex configurations, installations and maintenance, resulting in significant manpower savings and lower life cycle costs. With an open Service Orientated Architecture (SOA) using XML Medirva™ can utilize the latest developments in audio analytic tools thereby ensuring investment protection.

Medirva’s™ flexible and modular structure allows you to meet changing operational requirements and manage the resource allocation between the various audio analytic technologies and manual facilities.

By combining automatic analytic innovation, simplified methods for manual manipulation and comprehensive workflow management, Medirva™ reduces costs by ensuring that valuable resources are used efficiently with significantly improved operational capability. Further information on the applicability and usage of these technologies can be found in AudioSoft’s White Paper ‘Audio Mining – How to locate vital information automatically’.

Benefits

• Maximize Resources – Patented Resource Allocation
• Improved Capability – Automatically identify vital Intelligence
• Reduced Costs – Minimize manual resources requirements
• Manual Analysis – Noise filters, companders, expanders
• Understand relationships – Visualize results
• Ease of Use – Web based SOA architecture

Automated Capabilities

Speech Detection: Medirva’s™ Speech Detection algorithm automatically detects the location of speech even in noisy environments and is a mature technology ideally suited to surveillance operations where important speech may be buried within recordings that are primarily noise. It reduces manpower costs and avoids missing key sections of speech.

Speech to Text: Medirva’s™ Speech to Text solution provides speaker independent automated transcription of audio and requires no training. It is highly effective in situations where good quality audio must be transcribed, when a summary of subject content within a recording is required, or when combined with Contextual Search, to check thousands of hours of recordings for vital intelligence.

Contextual Search: Speech to Text is never going to be 100% accurate (neither is manual transcription). Medirva’s™ Contextual Search allows stems, synonyms and phonetic searches greatly improving the relevance of the speech to text process and identification of vital data. For instance ‘Heathrow’ may be automatically transcribed as ‘Heat’ and ‘row’ but will still be located by the contextual search engine.

Key Word Spotting: Medirva’s™ Key Word Spotting technology provides fast, automatic monitoring of thousands of hours of audio streams allowing audio material to be prioritized so that the most significant data can be handled first. It is used in situations where detection of specific words would warrant further detailed manual or automated analysis and is ideally suited to operations where large numbers of audio streams are being monitored.

Speaker ID: Medirva’s™ Speaker Identification solution distinguishes the speakers involved in a conversation extracts the characteristic voice print and compares the recorded speakers with voiceprints to identify the speakers. It is ideally suited to operations where the detection of a known target would lead to further more detailed manual or automated analysis of the audio data.

Language ID: Medirva’s™ Language Identification automatically segregates audio data into identified languages for either automated speech to text transcription using the appropriate language model or manual transcription by an appropriately skilled operator.

Auto gisting: Auto gisting produces an automatic summary of the content of an audio recording. It can then be used to prioritize further analysis and in situations where the detection of the most commonly used words or phrases is of relevance to an operation.

Visual Relationships: Medirva’s™ Visualization solutions allow users to quickly understand relationships between the recorded data. From timeline graphical displays of recording activity through to integrated map displays with GPS data to visual representation of complex scenarios our solutions allow our customers to better understand and act upon the recorded data. No more time consuming operations of sifting through data to establish relationships.

Manual Capabilities

Medirva™ provides a graphical display of the recorded data and allows the user to graphically manipulate the data through conventional tools such as zooming, panning to display regions of interest. Subtractive, band pass and adaptive filters can be applied to remove background noise whilst companders and expanders greatly aid the analyst in extracting vital information.

Workflow

Using our patented data interrogation technology Managers can automatically assign recorded data to the most appropriate automated or manual resources allowing valuable resources to be used efficiently whilst improving operational capability.