Architecture overview - AI Powered Speech Analytics for Amazon Connect

Architecture overview

Deploying this solution builds the following environment in the AWS Cloud.


        AI Powered Speech Analytics for Amazon Connect architecture on AWS

Figure 1: AI Powered Speech Analytics for Amazon Connect architecture on AWS

This solution extends your existing Amazon Connect environment by deploying the AWS services necessary to transcribe, translate, and analyze customer interactions.

  1. When a customer calls into your Amazon Connect call center, their call progresses through a contact flow. In the contact flow, the Start media streaming contact block captures customer audio. Then the Invoke AWS Lambda function contact block activates the invocation AWS Lambda function. As the call runs, the customer audio is streamed in real time to Amazon Kinesis Video Streams.

  2. The transcription Lambda function consumes the audio stream and uses Amazon Transcribe to convert the audio into text.

  3. The transcription Lambda function then stores the transcript segments and contact ID in an Amazon DynamoDB table. After the call ends, the captured transcripts will be uploaded to the solution-created Amazon Simple Storage Service (Amazon S3) bucket, with the Amazon S3 location added as an attribute to the Amazon Connect contact trace record. Contact center supervisors can mine the contact trace records for additional insights to improve the overall customer experience.

  4. When the call is routed to an agent, the agent’s call center application establishes a WebSocket connection to an Amazon API Gateway.

  5. The customer audio transcript is provided in real time to the agent. Additionally, Amazon Translate and Amazon Comprehend can provide translated and annotated transcripts, allowing the agent to efficiently find relevant information and perform recommended actions.

Note

AWS Lambda has a limitation where it stops audio processing when calls are longer than 15 minutes. To overcome this limitation, you can deploy an alternative setup. For more information, refer to Deploy an alternative CloudFormation stack.