Paper Title
Background Noise Reduction based on Wiener Suppression using a Priori Signal-to-Noise Ratio Estimate in Python

Random, additive noise is a form of degradation a major problem of all analog communication system. In an audio file, listeners can hear it as “hiss” sounds that usually came from different sources. These noises came from inside the device itself and ambient noise coming from the environment which is not correlated to the signal itself. This study focuses on the suppression of background noise in a mono-channel audio file using spectral analysis. Initially, the input audio file was segmented. Then, each segment was analyzed using Fast Fourier Transform and approximation of its SNR was calculated. The signal will then be attenuated by multiplying a suppression value calculated using the Wiener suppression formula in the frequency domain. The original signal was then restored using Inverse Fourier Transform and pass through a moving-average for smoothing. Data were gathered from actual noisy audio files which were the input to the system. The researchers compared the output audio signal to the input audio signal by evaluating the signal to noise ratio of the two. Index Terms - Background noise, audio segmentation, FFT, window, suppression rule, filter