"Integrated noise reduction algorithm"
Abstract
"Through the cascaded processing of four stages: adaptive notch filtering, multi-feature fusion Wiener filtering, 3D Kalman filtering, and lightweight residual convolutional network, collaborative enhancement from the time-frequency domain to the deep learning domain is achieved. Experiments show that this method can achieve an average signal-to-noise ratio improvement of over 12 dB in a 1-7 kHz interference environment, and its robustness and enhancement quality are significantly better than traditional single-domain algorithms."
Fields of Study & Subjects
Speech and Audio Processing
Advanced Adaptive Filtering Techniques
Image and Signal Denoising Methods
Publication details
- Journal / source
- IEEE DataPort
- Publication date
- 11 February 2026
- Language
- EN
- Document type
- Dataset