In recent years, the variety and volume of multimedia services have increased exponentially. Like most of other multimedia services, the conversational video service has stringent quality of service and experience requirements. In order to better support users QoE (Quality of Experience) and allocate network resources more effectively, this paper focuses on the QoS (Quality of Service)-QoE association modeling of conversational video flows. Through statistical analysis of the network video flow data, statistical distribution association between QoE and some network layer QoS parameters of the conversational video service is studied by use of the ITU-T P.1201.1 model, and relevant distribution functions are fitted. From the perspective of user’s experience, the QoS parameter recommendation indices are given for each MOS (Mean Opinion Score) value interval, and a QoE connected QoS model is established. Through the verification and comparison with an existing work, the rationality of the recommended QoS indices is demonstrated, which can help the network operators strike a balance between saving network resources and satisfying user’s quality of experience.