After conducting some preliminary research, reviewing the responses in this thread, and speaking with a few people, I came to the conclusion that the voice channel remains allocated for the entire AI interaction when using Genesys Cloud Audio Connector.
Audio Connector provides a WebSocket-based audio stream between Genesys Cloud and the AI platform, but it does not transfer ownership of the call or release the underlying voice resources.
As a result, the interaction continues to consume voice capacity within Genesys Cloud for as long as the call is active.
Thank you all for the responses!
------------------------------
Marcelo Breda
SOLUTION ARCHITECT SR
------------------------------