Processing Noisy Speech with Stationary Noise Using Spectral Subtraction Algorithm

In the MATLAB environment, implement spectral subtraction algorithm to process noisy speech containing stationary background noise. The algorithm works by estimating the noise spectrum during non-speech segments and subtracting it from the noisy speech spectrum in the frequency domain. Key implementation steps include voice activity detection, Fast Fourier Transform (FFT) analysis, and spectral magnitude adjustment with optional oversubtraction factor for better noise suppression. Please name your original speech file as "origin.wmv", and the processed output will be automatically saved as "enhanced.wmv". The RAR archive includes my pre-processed speech results for reference. You can replace these with your own audio files for processing by placing them in the same directory. Important: Always backup your original speech files before replacement to prevent data loss. The MATLAB code typically involves functions like audioread(), fft(), and spectral subtraction operations with proper overlap-add reconstruction for time-domain signal synthesis.