In order to process the audio data, we are using Audiohook to stream it. Additionally, we are loading our own application into the interaction widget and displaying the transcriptions within the application that is rendered inside the interaction widget.
However, the issue is that the call to audiohook is occurring even though the application in the interaction widget is not rendered.
Is it possible to confirm whether the interaction widget is loaded using the audiohook headers? or
Is there another way to avoid calling Audiohook on every call?
The audiohook session has no link/relation with an Interaction Widget which would be displayed in a Genesys Cloud Web UI. I mean that no information about the interaction widget is propagated to the audiohook session and vice versa.
What you could do is to have your Audiohook server automatically send a pause message when a new session is connected.
The websocket session must still be maintained but no audio is streamed to the audiohook server.
The id of the conversation is known to your server as it is propagated to the audiohook session on connect
When the call is received by the agent and when your webapp is displayed in the Interaction Widget, your webapp could contact your server and request to start the transcription (using the conversationId as the common identifier).
Your server could then send a resume message to request audio to be streamed again, and to start your voice transcription.