Hi Shaikha,
After extracting MFCC features, it's common to apply dimensionality reduction techniques such as PCA (Principal Component Analysis) and WPCA (Weighted Principal Component Analysis) to reduce the feature space and possibly improve classification or clustering performance.
- PCA is widely used for dimensionality reduction. In MATLAB, you can use the built-in pca function. Suppose your MFCC features are stored in a matrix called mfccFeatures, where each row corresponds to an audio sample and each column to an MFCC coefficient.
- WPCA is a variant of PCA where each sample can be assigned a weight, which is useful if you want certain samples to have more influence on the resulting components. While MATLAB does not have a built-in wpca function, you can implement it by weighting your centered data before applying PCA.
Documentation link : https://in.mathworks.com/help/stats/pca.html?requestedDomain=
