Answer Happy

1. Write your own Matlab/Python function for speech enhancement and test it with the file input.wav available to download from the I-Class website. For the short-time Fourier transform (STFT) analysis/synthesis, use the code that you wrote for HW2. 2. Save the enhanced speech signal as a .WAV file for online listening test. For Problem 1, write Matlab/Python function for speech enhancement, and write separate Matlab/Python codes to run this function. Submit the entire code and the enhanced output signal (in .WAV format) as a single zip file to I-Class by midnight on Tuesday June 14, 2022. 1 Problem 1 (15 points). Using any methods you have learned from the class (Wiener Fil- ter, Spectral Subtraction, MMSE-STSA methods, etc.) or any other methods including the deep neural network (DNN) that you can find, (or any algorithm you have invented), write a Matlab/Python code as a function to do speech enhancement. Please try various methods and only include the best Matlab/Python function that you wrote in your submission. You need to write a separate Matlab/Python code (not as a function) to load the noisy speech file (available from the I-Class webpage), run the speech enhancement function that you wrote, and write the output as a .WAV file. When you save the .WAV file, please make sure that the signal is normalized, i.e., with the maximum magnitude of 1. Please note that you need to estimate the noise power spectral density from the noisy input for this problem. Problem 2 (5 points). Please make sure that you include the output signal as .WAV file in your submission because we are going to do online subjective listening tests for the additional credit of 5 points. Other than including the WAV file of your output in your submission, you do not need to do anything further for Problem 2.