Veritone aiWARE 2.0 Accelerates and Customises AI Adoption

Veritone aiWARE2

Veritone aiWARE, one of the first operating systems for artificial intelligence, is now upgraded to version 2.0, making it easier to customise and use in a wider range of applications. aiWARE 2.0 has a real-time processing framework, greater cognitive capabilities, new support for electronic documents and structured and unstructured data, and new customisation options.

Veritone also has a suite of AI applications designed for specific uses, or users can create their own in-house applications that use aiWARE to drive processes and generate business insights.

AiWARE is a proprietary platform made for enterprises and government agencies to use as an extensible software infrastructure for developing AI deployments quickly and at scale. Users can implement, integrate and use aiWARE with their existing IT infrastructure, without extensive machine learning expertise. According to a 2019 survey by Gartner technological research company of CIOs interested in AI, these factors are among the greatest challenges for organisations adopting AI into their workflows.

Single Platform Architecture

aiWARE’s approach orchestrates a system of cognitive engines through a single platform architecture, giving users access to more than 320 AI engines that can be applied to audio, video and other data sources. These include Veritone’s proprietary engines, engines from cloud infrastructure providers, curated niche engines and others in order to take advantage of those resources with more scope, depth and accuracy than a single-AI-engine approach.

Because AI developments are emerging at a fast pace, Veritone considers that implementing any single system for handling data risks slowing the user down over time with outdated, limited capabilities. Most single point systems also cannot be expanded to add additional applications or cognitive capabilities. aiWARE’s platform avoids that risk by allowing users to add applications and cognitive engines from a wide pool of systems that is expanded and developed over time.

Veritone aiWARE Platform Architecture

More Engines, More Customisation

Real-Time Framework

aiWARE’s real-time framework now supports all of Veritone’s cognitive capabilities, which address face recognition, object detection, content classification, image classification and many others. They are what the platform uses to support use cases that require real-time data insight extraction or cognitive workflows – that is, those that determine the flow of analysis by events such as user actions, sensor outputs or messages from other programs.

Seeing results within a few seconds, users can make decisions faster and simplify their workflows to improve the efficiency and effectiveness of operations. For example, processing broadcast audio through aiWARE’s transcription engines immediately delivers text that can be used for closed captioning.

Data Types

aiWARE 2.0 has expanded support four different data types and formats, such as Word and PDF files, which means it can now also use both structured and unstructured data to generate insights from complex analyses. Users can use Veritone applications or developers to create their own applications that analyse and correlate sources such as public and private databases with unstructured files - photos, video and audio.

Using structured data, for example, applications such as Veritone IDentify can not only identify individuals using facial recognition but can also correlate identified people with records, reports and related information to speed up investigations. In the media and entertainment industry, Veritone Attribute uses this new structured data capability to ingest and analyse an advertiser’s website traffic performance measured by Google Analytics, and correlate that data with live audio and video broadcast streams to create attribution models.

Veritone aiWARE

AI-Enabled Applications

aiWARE 2.0 now includes six AI-enabled business applications that Veritone developed to tackle specific industry challenges, without requiring AI expertise. As well as IDentify and Attribute, these applications are Veritone Redact, Veritone Core, Veritone Digital Media Hub and Veritone Commerce.

aiWARE 2.0 includes Veritone’s machine learning tools that allow users to create customised engines across multiple cognitive capabilities. Whether to create new or enhance existing business systems, these proprietary Veritone engines can be trained for specific use cases to improve accuracy.

Veritone’s most recent engines can be deployed as a part of a cloud or hybrid system within aiWARE to conform with security and regulatory compliance requirements. For example, a major broadcaster might need to track appearances of a new brand logo across its television broadcasting. Its staff can train a customised logo-detection model and then deploy it into the aiWARE environment for real-time brand identification.

Expanding Platform

The Veritone platform has grown, and today includes more than 320 engines across several new cognitive capabilities, such as speaker separation, with more capabilities and engines being continuously added. Individually and working in combinations, these cognitive capabilities further expand the range of potential use cases for Veritone.

To illustrate speaker separation, law enforcement agencies commonly record conversations in prisons. However, these conversations often shift between multiple languages, making transcription very difficult. aiWARE can assist agencies by employing more than one natural language processing engine to transcribe and translate these recorded conversations.

New self-service tools in aiWARE 2.0 make it easier for enterprises to use Veritone’s configurable platform architecture to create systems customised to the type of work they do. Greater flexibility in implementing new cognitive engines, new data sources and new data schemes makes the Veritone platform a longer term investment.