A cough sound recognition method based on a PSO-GBDT-LR model

CN120388587BActive Publication Date: 2026-06-26KUNMING UNIV OF SCI & TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
KUNMING UNIV OF SCI & TECH
Filing Date
2025-05-22
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing cough detection methods rely on doctors' subjective judgment, leading to inconsistent assessment results. Furthermore, noise interference and individual differences exist in cough audio signal recognition, making it difficult to accurately distinguish between cough and non-cough sounds.

Method used

A cough sound recognition method based on the PSO-GBDT-LR model is adopted. The method uses Berouti spectrum subtraction for noise reduction, audio event detection, extraction of 9-dimensional feature vectors, and particle swarm optimization algorithm to adjust hyperparameters. The GBDT-LR model is then used for classification.

Benefits of technology

It improves the accuracy and generalization ability of cough sound recognition, reduces noise interference, reduces computational load, avoids complex feature extraction processes, and achieves accurate differentiation between cough and non-cough sounds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120388587B_ABST
    Figure CN120388587B_ABST
Patent Text Reader

Abstract

The application discloses a cough sound recognition method based on a PSO-GBDT-LR model and belongs to the technical field of signal recognition. The method comprises the following steps: collecting an audio signal; performing denoising on the audio signal by using a Berouti spectrum subtraction method to obtain a denoised audio signal; performing audio event detection VAD to segment the part where the audio appears sound; extracting 7-dimensional time domain features for each segmented audio sample; performing short-time Fourier transform STFT on each segmented audio sample to extract 2-dimensional frequency domain features from the frequency spectrum; combining the extracted 7-dimensional time domain features and the extracted 2-dimensional frequency domain features to form a 9-dimensional feature vector combination; marking the feature vector of the cough audio sample as class 1, marking the feature vector of the non-cough audio sample sound sample as class 0 and the like. The application solves the problems of too many noise features and abnormal features of sound in the cough sound recognition process, can accurately distinguish the features of cough sound and non-cough sound, and has strong generalization ability.
Need to check novelty before this filing date? Find Prior Art