Calls made to a business by a customer, e.g., to request support, are often put into a queue waiting for a human agent to be available. During the hold time, canned music or other audio is played back to the caller. Such audio is low quality owing to the limited capacity of the telephony channel, is not personalized, and repeated multiple times till an agent becomes available, providing an unsatisfactory calling experience. This disclosure describes the use of machine learning techniques to detect canned audio and replace it with high fidelity music or other content. With user permission, the replacement content can be personalized, e.g., based on a user’s music playlists/preferences, and context. Machine learning techniques can also be utilized to upscale music on hold experience provided by the business. With user permission, advertising content or helpful content about the business can be delivered during the hold time. The techniques can be integrated into a virtual assistant or device operating system to provide an improved calling experience.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.