Non-Realtime Dual-Channel Audio Processing for Echo Cancellation

This repository contains a MATLAB script (EchoCanceller.m) for non-realtime audio processing, focusing on echo cancellation for dual-channel audio files. The script processes WAV files with near-end microphone and far-end speaker signals, applying frequency-domain NLMS-based echo cancellation to produce a clean output with reduced echo.

Features

• Echo Cancellation: Implements frequency-domain NLMS-based echo cancellation with adaptive spectral smoothing to effectively suppress echo in dual-channel audio.
• Dual-Channel Support: Processes dual-channel WAV files (near-end microphone and far-end speaker signals).
• Dynamic Frequency Range: Adjusts echo frequency range (100 Hz to 4000 Hz) to cover low-frequency echo and speech frequencies.
• Sampling Rate Support: Supports 8 kHz and 16 kHz sampling rates with adjustable parameters (cohRange, mufb).
• Post-Processing: Includes residual echo suppression and dual-channel gain normalization to maintain energy levels.
• Visualization: Provides detailed plots for analysis, including time-domain waveforms, spectrograms, and energy (dB) plots for both input and output signals.
• Energy Compensation: Ensures energy retention through dual-channel gain normalization, preserving natural speech levels.

Prerequisites

• MATLAB: This script requires MATLAB (tested with R2020a and later versions).
• Audio File: A dual-channel WAV file with the following specifications:
  • Channel 1: Near-end microphone signal (with echo).
  • Channel 2: Far-end speaker signal (reference for echo cancellation).
  • Supported sampling rates: 8 kHz or 16 kHz.

Installation

1. Clone the repository to your local machine:

   git clone https://github.com/MetaAcoustix/EchoCanceller.git

2. Navigate to the repository directory:

   cd EchoCanceller

3. Ensure MATLAB is installed and added to your system path.

Usage

1. Prepare Input File:

   • Place your dual-channel WAV file (e.g., original_signal.wav) in the repository directory.
   • Ensure the file meets the requirements (dual-channel, 8 kHz or 16 kHz sampling rate).

2. Run the Script:

   • Open MATLAB and navigate to the repository directory.
   • Open the script EchoCanceller.m.
   • Update the filename variable in the script to match your input file:

     filename = 'original_signal.wav';

   • Run the script:

     run EchoCanceller.m

3. Output:

   • The processed audio will be saved as mout_core_original_signal.wav in the same directory.
   • Two figures will be displayed:
     • Input Signal Acoustic Plot: Time-domain waveforms, spectrogram, and energy plot for the input near-end signal.
     • Output Signal Acoustic Plot: Time-domain waveforms, spectrogram, and energy plot for the processed near-end signal.

4. Analyze Results:

   • Check the output WAV file for echo suppression and speech clarity.
   • Review the spectrogram to confirm echo reduction.
   • Inspect the energy plot to ensure energy levels are preserved.

Example

Input File

• File: original_signal.wav
• Format: Dual-channel WAV, 16 kHz sampling rate
• Channel 1: Near-end microphone signal (with echo)
• Channel 2: Far-end speaker signal

Run Command

run EchoCanceller.m

Output File

• File: mout_core_original_signal.wav
• Format: Dual-channel WAV, 16 kHz sampling rate
• Channel 1: Processed near-end signal (echo suppressed)
• Channel 2: Original far-end signal (unchanged)

Visualization

• Input Spectrogram: Shows echo and noise in the near-end signal.
• Output Spectrogram: Shows reduced echo with preserved speech.

Testing

To test the script:

1. Use a dual-channel WAV file with known echo (e.g., recorded during a speakerphone call).
2. Run the script as described in the Usage section.
3. Verify the output:
   • Listen to mout_core_original_signal.wav to confirm echo reduction.
   • Check the spectrogram for echo suppression (reduced energy in echo frequency bands).
   • Ensure speech clarity is maintained and no significant distortion is introduced.

Contributing

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a new branch for your feature or bug fix:

   git checkout -b feature/your-feature-name

3. Make your changes and test thoroughly.
4. Commit your changes with a descriptive message:

   git commit -m "Add feature: your feature description"

5. Push to your fork:

   git push origin feature/your-feature-name

6. Open a Pull Request with a detailed description of your changes.

Potential Improvements

• Add support for additional sampling rates (e.g., 44.1 kHz).
• Integrate noise suppression for specific frequencies (e.g., 440 Hz).
• Enhance distortion handling with low-pass filtering.
• Add non-linear processing (NLP) for residual echo suppression.

License

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