Hearing Loss Emulation and Denoising Plugin

Overview

The Hearing Loss Emulation Plugin and Denoising Script is a MATLAB-based tool designed to simulate hearing loss and demonstrate the need for hearing aids. This project uses various acoustic parameters to model hearing impairment and provides an FFT-based denoising script to simulate how hearing aids can reduce unwanted noise and enhance important sounds like speech.

Purpose

This project aims to:

• Emulate Hearing Loss: Simulate various types of hearing loss using customizable parameters to reflect the impact on auditory perception.
• Justify the Need for Hearing Aids: Use the simulated hearing loss to highlight how hearing aids can improve sound quality and clarity for individuals with hearing impairment.
• Denoise Audio: Implement FFT-based denoising to illustrate how hearing aids reduce unwanted noise and focus on important sounds.

Features

• Hearing Loss Simulation: Adjust high, mid, and low frequencies, apply filters, and simulate spectral smearing to model different aspects of hearing loss.
• Noise Injection: Add simulated noise to reflect conditions like tinnitus or background noise.
• FFT-Based Denoising Script: A MATLAB script that uses Fast Fourier Transform (FFT) to reduce noise and enhance speech signals, simulating the noise reduction capabilities of hearing aids.

Installation

1. Prerequisites: Ensure you have MATLAB installed with the Audio Toolbox.

2. Clone the Repository:

   git clone https://github.com/calanuzao/hearingaidEQ.git

3. Navigate to the Project Directory:

   cd hearingaidEQ

4. Add the Plugin to MATLAB:

   • Open MATLAB.
   • Navigate to the directory containing the hearinglosseq.m file.
   • Add the directory to the MATLAB path:

     addpath('path/to/hearingaidEQ')

Usage

Hearing Loss Emulation

1. Configure Parameters:

   plugin.HS_FREQ = 3000;     % Set high-frequency shelf frequency
   plugin.HS_GAIN = -30;      % Set high-frequency shelf gain
   plugin.HMF_FREQ = 1000;    % Set mid-frequency
   plugin.HMF_GAIN = -10;     % Set mid-frequency gain
   plugin.LMF_FREQ = 500;     % Set low-mid frequency
   plugin.LMF_GAIN = -10;     % Set low-mid frequency gain
   plugin.HPF_FREQ = 30;      % Set high-pass filter frequency
   plugin.LPF_FREQ = 1500;    % Set low-pass filter frequency
   plugin.LPF_GAIN = -6;      % Set low-pass filter gain
   plugin.SMEAR_FREQ = 1500;  % Set spectral smearing frequency
   plugin.SMEAR_GAIN = -20;   % Set spectral smearing gain
   plugin.NOISE = 'on';       % Enable noise
   plugin.NOISE_GAIN = 0.011; % Set noise gain
   plugin.BYPASS = 'off';     % Disable bypass

2. Run Plugin through the MATLAB Command Window or as a VST/VST3/AU Plugin on your machine.

3. Add Input Audio Files (.wav or .mp3).

4. Reset Plugin.

FFT-Based Denoising Script

To use the FFT-based denoising script, follow these steps:

1. Open the Script:

   • Locate the denoisingdata.m script in the project directory.

2. Run the Script:

   % Example usage of the FFT-based denoising script
   % Generate a noisy signal
   dt = 0.001; % Time step
   t = 0:dt:1; % Time vector
   fclean = sin(2*pi*50*t) + sin(2*pi*120*t); % Clean signal
   f = fclean + 2.5*randn(size(t)); % Add noise
   
   % Plot the time-domain signal and its frequency domain representation
   figure; set(gcf, 'Position', [1500 200 2000 2000])
   subplot(3,1,1)
   plot(t, f, 'c', 'LineWidth', 3); hold on
   plot(t, fclean, 'k', 'LineWidth', 2.5)
   l1 = legend('Noisy', 'Clean'); set(l1, 'FontSize', 14)
   xlabel('Time (s)', 'FontSize', 14)
   ylabel('Amplitude', 'FontSize', 14)
   title('Time-Domain Signal', 'FontSize', 16)
   ylim([-10 10]); set(gca, 'FontSize', 14)
   
   % FFT
   n = length(t); 
   fhat = fft(f, n); % Compute FFT
   PSD = abs(fhat).^2 / n; % Power spectrum
   freq = (0:n-1) * (1/(dt*n)); % Frequency vector
   L = 1:floor(n/2); % Only plot the first half of frequencies
   
   % Plot the PSD before filtering
   subplot(3,1,2);
   plot(freq(L), PSD(L), 'c', 'LineWidth', 3); hold on
   xlabel('Frequency (Hz)', 'FontSize', 14)
   ylabel('Power/Frequency (dB/Hz)', 'FontSize', 14)
   title('Power Spectral Density Before Filtering', 'FontSize', 16)
   set(gca, 'FontSize', 14)
   
   % Filter noise using PSD
   indices = PSD > 100; % Threshold for filtering
   PSDclean = PSD .* indices; % Apply filter
   fhat = fhat .* indices'; % Zero out small coefficients
   ffilt = ifft(fhat); % Inverse FFT for filtered signal
   
   % Plot the PSD after filtering
   subplot(3,1,3);
   plot(freq(L), PSD(L), 'c', 'LineWidth', 3); hold on
   plot(freq(L), PSDclean(L), '-', 'Color', [.5 .1 0], 'LineWidth', 2.5)
   xlabel('Frequency (Hz)', 'FontSize', 14)
   ylabel('Power/Frequency (dB/Hz)', 'FontSize', 14)
   title('Power Spectral Density After Filtering', 'FontSize', 16)
   legend('Original', 'Filtered', 'FontSize', 14)
   ylim([-10 10]); set(gca, 'FontSize', 14)

3. Analyze Results:

   • Review the time-domain and frequency-domain plots to compare the noisy and denoised signals.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

• MATLAB Audio Toolbox for providing essential tools for audio processing.
• GitHub for hosting the project repository.
• NYU for providing the COMSOL Multiphysics Acoustic Modules and MATLAB License.

CALODII STUDIOS Copyright 2024