This repository contains code to the research project Efficient measurement matrix for speech compressive sampling. In this work, an efficient sensing matrix is designed using genetic algorithm and learning based method in order to replace the traditionally used random matrices such as bernouli, gaussian or hadamard. A comparative analysis is also demonstrated where the designed sensing matrix outperforms other sensing matrices in achieving the lowest reconstruction error and high Signal to Noise ratio.
- Scipy
- Numpy
- CVX
- Sklearn
- Python.
- Jupyter Notebooks.
- MATLAB.
The research article published based on the presented work can be accessed at the following link.
https://link.springer.com/article/10.1007/s11042-021-10657-x