Appendix A: Controlling the Solution Behavior - AI Powered Speech Analytics for Amazon Connect

Appendix A: Controlling the Solution Behavior

The AI Powered Speech Analytics for Amazon Connect solution has flags you can use in the kvsConsumerTrigger AWS Lambda function which will then pass the attributes to the kvsTranscriber Lambda function. The following flags will control the behavior of the solution:

Lambda function Default Description
transcribeCall true Enable call transcriptions
saveCallRecording false Choose whether to save the audio captured from the live audio stream to the customer-defined Amazon S3 bucket
languageCode en-US or es-US The language the customer selected when calling in, and transcribed to the agent using Amazon Transcribe. For more information on supported languages and codes, see Streaming Transcription in the Amazon Transcribe Develper Guide.

The kvsTranscriber Lambda function has the following flags that can be modified:

Lambda function Default Description
CONSOLE_LOG_TRANSCRIPT_FLAG TRUE Enable call logging for transcripts in CloudWatch
SAVE_PARTIAL_TRANSCRIPTS TRUE Logs and saves all partial transcripts from Amazon Transcribe. If you set this to false you will only see the final result returned from Amazon Transcribe.
START_SELECTOR_TYPE NOW Used to tell Transcribe when to start the transcription. Select NOW to dstart transcription once the agent and customer are connected. Selecgt FRAGMENT_NUMBER to start transcription once the Start Media Streaming block is invoked in the contact flow.

Note that this solution includes an agentAssist.html file that contains a mapping of key phrases to suggested actions. You can customize this file to fit your needs. For example, the file contains a mapping that suggest the action to "offer discount" if Amazon Comprehend detects the key phrase "cancel my service".