A sound source identification method based on array expansion and SE-moving inverse bottleneck convolutional neural network
By combining EAG-U-Net and SE-MBCNet networks, the problem of insufficient accuracy and robustness of traditional sound source localization methods in complex environments is solved, and the effect of improving the accuracy and clarity of sound source localization is achieved without increasing the physical array.
Patent Information
- Application Number
- CN202411877044.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-19
- Publication Date
- 2025-11-14
- Estimated Expiration
- 2044-12-19
AI Technical Summary
Traditional sound source localization methods suffer from reduced accuracy and robustness in complex environments with high noise levels or a limited number of arrays. Furthermore, increasing the number of sensors leads to higher hardware costs and computational burdens, making it difficult to improve localization accuracy without increasing the physical array.
A sound source identification method based on array extension and SE moving inverse bottleneck convolutional neural network is adopted. The sound source distribution map predicted by the 18-array MUSIC algorithm is converted into the sound source distribution map predicted by the 64-array MUSIC algorithm through the EAG-U-Net data conversion model. The image processing is combined with the SE-MBCNet network to improve the localization accuracy and robustness.
It significantly improves the accuracy and robustness of sound source localization, reduces the number of microphone arrays required, enhances its application capabilities in complex acoustic environments, and improves the clarity and accuracy of sound source distribution maps.