Computer implementation methods, computer programs, and systems (automated domain-specific constraint decoding from voice input to structured resources)

Automated domain-specific constraint decoding with a trained AI engine addresses inaccuracies in conventional speech recognition, providing precise text and action outputs for structured resources.

JP7868938B2Active Publication Date: 2026-06-02INTERNATIONAL BUSINESS MACHINE CORPORATION
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
INTERNATIONAL BUSINESS MACHINE CORPORATION
Filing Date
2022-12-19
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Conventional speech recognition methods often produce inaccurate and/or incomplete outputs.

Method used

Implementing automated domain-specific constraint decoding using an artificial intelligence-based decoding engine trained with a domain-specific model, which processes speech utterances to generate accurate domain-specific text and action outputs for structured resources.

Benefits of technology

Enhances the accuracy and completeness of speech recognition by constraining the decoding process with domain-specific information, ensuring correct selection of form elements and insertion of text into structured resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007868938000001
    Figure 0007868938000001
  • Figure 0007868938000002
    Figure 0007868938000002
  • Figure 0007868938000003
    Figure 0007868938000003
Patent Text Reader

Abstract

To provide a method, a system and a computer program for automated domain-specific constrained decoding from speech input to a structured resource.SOLUTION: A method includes steps of: converting at least a part of speech utterance provided by at least one user into text by processing the speech utterance provided by at least one user by using an artificial intelligence-based automatic speech recognition model; automatically training an artificial intelligence-based decoding engine, which comprises a step of constraining the artificial intelligence-based decoding engine based on at least partially a domain-specific model and the artificial intelligence-based automatic speech recognition model; and generating at least one of domain-specific text output related to a structured resource associated with a domain and domain-specific action output by processing at least a part of the text by using the artificial intelligence-based decoding engine.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art