A seamless fallback method, system, device, and medium for streaming speech translation

By building an audio buffer queue and multi-level degradation links on the server side, seamless degradation of the voice translation system was achieved, solving the problems of data loss and user experience disruption caused by underlying network link failures, ensuring the continuity and integrity of recognition results, and improving the system's disaster recovery capabilities.

CN122245317APending Publication Date: 2026-06-19SHENZHEN XINZHILIAN SOFTWARE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHENZHEN XINZHILIAN SOFTWARE CO LTD
Filing Date
2026-05-21
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

In real-time speech translation and simultaneous interpretation scenarios, existing technologies suffer from audio data loss and user experience disruption due to underlying network link failures. Traditional disaster recovery solutions cannot achieve seamless switching and data inheritance, affecting the continuity and integrity of recognition results.

Method used

On the server side, an audio buffer queue independent of specific speech recognition and translation engines is built, and a multi-level degradation link is constructed. When an engine fails, a state transition algorithm is used to switch to a backup engine, while retaining the core metadata and context information of the session, so as to achieve seamless switching and data inheritance.

Benefits of technology

It solves the problem of audio data loss, ensures the continuity and integrity of recognition results, improves the system's disaster recovery capability and user experience, avoids repeated recognition or recognition interruption, and achieves seamless degradation of speech recognition and translation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122245317A_ABST
    Figure CN122245317A_ABST
Patent Text Reader

Abstract

This invention discloses a seamless degradation method for streaming speech translation, comprising the following steps: constructing an audio buffer queue independent of the speech recognition engine and translation engine on the server side, and appending all received audio streams to the audio buffer queue; constructing a speech recognition degradation link containing at least two speech recognition engines, and a translation degradation link containing at least two translation engines; if the current speech recognition engine fails, starting the next priority speech recognition engine, and migrating the session core metadata of the audio buffer queue to the new engine through a state transition algorithm, allowing the new engine to continue recognition from the queue breakpoint; if the current translation engine fails, retaining the original text and context information of the current translation task, starting the next priority translation engine, and injecting the original text and context information into the new engine for silent retry. This achieves seamless switching in the event of underlying service failures.
Need to check novelty before this filing date? Find Prior Art