An AI-Based Framework for Speech and Voice Analytics to Automatically Assess the Quality of Service Conversations

In this chapter, an innovative two-stage classification framework is presented that can predict quality-inducing criteria in call center conversations with explainable rules based on multiple models for speech expression. Through this basic classification, a symbolic representation of the speech expression is generated that is both understandable to experts and can be processed by classification algorithms. In the second stage, learning procedures are used to combine the recognized speech and voice features into a classification of quality factors. Rules and decision trees map the functional relationships to the relevant features and can thus explain the perceived quality factors on the basis of the recognized speech-voice features.