Voice interaction method, device and electronic equipment

By adopting an end-to-end single-model architecture and combining speech coding and text coding features, the problem of intent recognition accuracy in existing SLU systems when faced with incomplete or ambiguous speech input is solved, achieving efficient and accurate intent recognition and multi-turn dialogue clarification, and improving the overall performance of voice interaction.

CN121075331BActive Publication Date: 2026-05-22HANGZHOU QIUGUOJIHUA TECHNOLOGY CO LTD
3 Cites 0 Cited by

Patent Information

Application Number
CN202511632682.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-11-10
Publication Date
2026-05-22
Estimated Expiration
2045-11-10

AI Technical Summary

Technical Problem

Existing SLU systems struggle to effectively clarify incomplete, ambiguous, or ambiguous voice input from users through multiple rounds of dialogue, resulting in low accuracy in intent recognition and failing to meet the needs of complex human-computer interaction scenarios.

Method used

It adopts an end-to-end single-model architecture, which acquires user input voice information and historical intent text, generates acoustic coding features and text coding features using pre-trained voice encoders and text encoders, and combines them with an intent recognition module to perform intent recognition. It abandons the traditional cascaded scheme and uses a multi-turn dialogue mechanism to improve accuracy.

Benefits of technology

It significantly improves the accuracy of intent recognition results, quickly and accurately obtains the user's true intent, shortens the processing link from voice input to intent output, and reduces the impact of ASR module errors on NLU module.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application provides a speech interaction method and device and electronic equipment, and relates to the technical field of speech processing. The method comprises the following steps: acquiring speech information input by a user, and acquiring historical intention text of the user; inputting the speech information into a speech encoder of a spoken language understanding model to obtain acoustic coding features output by the speech encoder; inputting the historical intention text into a text encoder of the spoken language understanding model to obtain text coding features output by the text encoder; inputting the acoustic coding features and the text coding features into an intention recognition module of the spoken language understanding model to obtain an intention recognition result output by the intention recognition module, so as to be used for speech interaction. The application can improve the accuracy of the intention recognition result and accurately acquire the real intention of the user.
Need to check novelty before this filing date? Find Prior Art