Automatic Voice Quality Evaluation Method of lVR Service in Call Center Based on Stacked Auto Encoder

The application of virtual reality service in call center makes the voice evaluation quality of call center a research hotspot. Previous voice quality assessment methods could not solve the problem of stacked encoding in call centers, and the voice quality assessment capability was low. Therefore, this paper proposes a virtual reality method to construct a voice quality evaluation system for call centers. First, stacked autoencoders are used to classify the data of the call center and divide the collection according to the department, which reduces the complexity of data processing. The autoencoder then categorizes the call center into a collection of voice quality, and the data in the collection is automatically evaluated. In MATLAB simulation, the evaluation accuracy and calculation time of the virtual reality method are better than the previous speech evaluation methods under the condition that the amount of call data is fixed.