Oral language learning methods, devices, equipment and storage media

Abstract

This application discloses a spoken language learning method, apparatus, device, and storage medium. It supports user-defined spoken dialogue scenarios, obtains user-defined scenario description information, and generates and outputs the machine's current round of dialogue speech based on this scenario description information, the user's current input speech, and the user's previous historical dialogue speech before the current round. This achieves the goal of user-machine spoken language practice within user-defined scenarios during spoken language learning. This application abandons the traditional approach of generating machine response speech through user speech recognition, machine response text generation, and speech synthesis, achieving end-to-end speech generation. It preserves the prosodic information of the user's speech in the historical human-machine dialogue, making the generated machine response more consistent with the user's current emotional state, increasing the user's willingness to engage in spoken dialogue with the machine, and improving the efficiency of user spoken language learning.

Images