Amazon Web Services Inc. announced today that it’s enabling call analytics for the Amazon Chime software development kit, making it easier and more cost-effective to generate insights on real-time audio calls.

Amazon Chime SDK call analytics enables developers to create transcriptions of real-time calls, perform voice tone analysis and search for specific speakers, the company said. In addition, AWS is updating the Amazon Chime SDK section of the AWS Management Console to make it easier to integrate these new capabilities within audio applications.

Amazon Chime SDK is a set of real-time communications components for developers that make it easy to add messaging, audio, video and screen-sharing capabilities to web and mobile applications. For example, developers can use it to add video to a health app, enabling patients to enter into a video call with their doctor directly through the health provider’s application.

In a blog post, Amazon principal developer advocate Sébastien Stormacq explained that there are situations where customers may want to record and analyze calls made through their audio applications. The new voice analytics feature is designed to generate real-time insights based on these calls by using machine learning to extract sentiment from speech signals in order to detect things like dissatisfaction or impatience. It works by performing a joint analysis of lexical and linguistic information — namely, what was said — and combining this with acoustic and tonal information — how it was said.

According to Stormacq, these voice tone analytics will be delivered to the data lake of the customer’s choosing, allowing them to create dashboards that can visualize this data.

Amazon provided an example of how this might work in the finance industry, where trading room supervisors often record all conversations that take place, for compliance reasons. Voice tone analysis will help supervisors to monitor for risk and compliance, Stormacq said. In addition, the insights can help traders to improve their productivity.

Other industries where voice analysis might be helpful include healthcare, public sector, telecommunications and insurance.

The new speaker search capability is designed to detect and recognize speakers on a call. It can identify individual speakers after listening to only a few seconds of their speech, based on their voice which is stored in a database of all known voices. Stormacq said that can help with tasks such as expediting caller lookup and enriching transcripts of calls with identity attribution.

To make it easier for developers to use these new capabilities, Amazon has updated the AWS Management Console. Developers can simply choose the AWS artificial intelligence service they’d like to use to analyze a specific call, such as voice analytics, Amazon Transcribe or Amazon Transcribe Call Analytics.

To visualize these analytics, users can choose from Amazon QuickSight or Tableau, whose dashboards can then be embedded into a variety of applications, wikis or portals, Amazon said. And rather than create their own dashboards, developers can download prebuilt ones as AWS CloudFormation templates.

Finally, call analytics can be used to generate real-time alerts by posting specific events to Amazon EventBridge, which then routes these notifications to the user’s AWS account or any supported third-party application.

Liz Miller of Constellation Research Inc. told SiliconANGLE that developers will welcome the new call analytics capabilities in Chime because this kind voice and sentiment data is vital for training new AI and machine learning models.

“This latest SDK gives Chime and AWS customers exactly what they need to  turn voice into action,” Miller said. “While AI and ML have been announced by almost every major contact center provider, the interesting part of the AWS announcement is that this is really about providing teams building tomorrow’s experiences with the developer toolkits they need to accelerate time to value.”

Amazon said pricing for Amazon Chime SDK call analytics will be based on usage, with users charged per minute of audio data analyzed. Amazon Chime SDK call analytics is currently available in AWS’s US East (Ohio, N. Virginia), Asia Pacific (Singapore) and Europe (Frankfurt) regions, with more to come later.

Images: AWS