Content by Sheng Liu, Justin Gonzalez, Shivank Goel, Morteza Ziyadi (1)
Sheng Liu, Justin Gonzalez, Shivank Goel, and Morteza Ziyadi introduce four Microsoft Foundry evaluators that assess voice-agent conversations directly from recordings. They explain the required audio-capable judge configuration and report benchmark accuracy, repeatability, and satisfaction-score agreement, while noting that results depend on the tested dataset and setup.
End of content