Configuring the Voice AI configuration set
Configure the Voice AI configuration set to enable your application to use Voice AI and to define high-level configuration for the Voice AI solution.
Pega Customer Service
Implementation Guide
- In the navigation pane of App Studio, click .
- In the Configuration set: Voice AI section, use the
Settings icon for each property to configure the
following properties:
- Entity expiration
- The number of seconds after which a detected entity becomes stale and is not used for form autofill.
- The default value of 0 means that the entity never becomes stale. Enter a custom value depending on your business requirements.
- Isolation ID
- The unique isolation ID of your system.
- You can construct your isolation ID by appending the type of environment to your AWS Cognito client ID (typically known as the Cognito ID) that Pega provides to you during onboarding. For example, <CognitoID>dev1, <CognitoID>stg3, or <CognitoID>prod2. Acceptable characters in an isolation ID are A-z, 0-9, ., -, _, and ~. If you need assistance with your Cognito ID, contact your Pega representative.
- Language Model
- Select General.
- Additional industry-specific language models might be available. For more information, contact your Pega representative.
- Persist Audio Transcript
- Enter True to save transcripts to the database so that CSRs and supervisors can review and verify the transcripts.
- Speech Model
- Select en_US.
- Additional speech models might be available. For more information, contact your Pega representative.
- Transcript resource
- The resource path parameter of the REST Connect URL that the system uses to retrieve the transcripts.
- Enter transcripts.
- Transcript version
- Enter v1.
- VAD Record Only Mode
- Select False.
- As of Pega Customer Service version 8.6, the VAD record only mode property is a placeholder for future implementation. In future versions, you can select True to disable transcription and record the audio for training purposes only. Doing so stores the audio streams in an external storage location without any speech processing, and the CSR does not receive the transcription.
- Verbatim dialog threshold
- The threshold that the natural language processing (NLP) engine requires to consider the words that are spoken by an agent as a match to a dialog that requires a verbatim match in the case configuration.
- Setting the value to 100 can cause normal variations in speech patterns, resulting from pauses, filters, and dialects, to create a large number of mismatches. Retain the default value of 80 for optimal performance.
- Voice AI Channel ID
- Your Voice AI channel ID, which you can find on the Behavior tab of your Voice AI channel landing page.
Previous topic Preparing for Voice AI implementation Next topic Setting up the Voice AI authentication profile