A content-aided synthetic audio detection method

By constructing a semantic-acoustic dual-path joint analysis architecture, combining natural language processing and acoustic features, the detection difficulties of existing technologies in multi-speaker aliasing scenarios are solved, achieving accurate detection and intent classification of synthesized audio, and improving the adaptability and accuracy of the detection system.

CN121922154BActive Publication Date: 2026-07-03LANZHOU UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
LANZHOU UNIV
Filing Date
2025-10-24
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing synthetic audio detection methods suffer from performance degradation in multi-speaker aliasing, cross-dialogue, and complex audio scenarios. They struggle to accurately classify the intent categories of synthesized speech and lack adaptability to real-world scenarios, resulting in high false positive and false negative rates.

Method used

A semantic-acoustic dual-path joint analysis architecture is constructed. By combining natural language processing and acoustic feature extraction with a multi-speaker aliased speech dataset to train the model, accurate detection and intent classification of synthesized audio are achieved.

Benefits of technology

It significantly improves the accuracy and robustness of synthetic audio detection, reduces false alarm and false negative rates, and enables fine-grained risk classification and collaborative alarm in complex scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121922154B_ABST
    Figure CN121922154B_ABST
Patent Text Reader

Abstract

This invention discloses a content-assisted synthetic audio detection method. It extracts acoustic and natural language semantic features in parallel for each speech segment, determines the level of co-alarm based on the two types of features, and adjusts the level of co-alarm according to the text intent and acoustic style while detecting synthetic speech. Finally, a structured analysis report is generated. By deeply introducing natural language processing methods into the synthetic audio detection process, a semantic-acoustic dual-path joint analysis architecture is constructed, which breaks through the limitations of traditional detection schemes that rely only on acoustic fingerprints, spectral features or model generation traces.
Need to check novelty before this filing date? Find Prior Art