Call analysis model updates

Genesys Cloud continuously improves call analysis model updates on a monthly basis to enhance detection accuracy of carrier tones and spoken carrier messages across different countries. The system uses audioprints (fingerprints) to detect carrier messages such as SIT-related, busy, and machine messages with high accuracy through a model update process. For optimal fingerprinting results when reporting misdetected messages, recordings should be captured with dual channel enabled on the trunk, which stores caller and agent audio in separate streams and allows the service team to extract clearer, more complete audioprints from the caller's audio. Without dual channel, both parties are captured in a single combined recording with overlapping audio, potentially yielding only partial fingerprints. For premises-based deployments, all LDM Edges across the organization must be upgraded to the latest version to benefit from model updates; otherwise, the model associated with the oldest Edge version is used for all Edges regardless of individual Edge versions. Cloud media CDM Edges do not require manual updates, as model updates are delivered through cloud media services. If carrier messages or tones are misdetected during Outbound dialing, customers should contact Customer Care. Supported call analysis regions are available for reference to identify where these models are deployed.