Hi,
To boost classification accuracy, you can refine your feature set using PCA or LDA. Normalize your data and reduce noise for cleaner inputs. Experiment with different algorithms such as SVM, k-NN, or decision trees, and fine-tune them using cross-validation. If class imbalance is an issue, try balancing methods like oversampling or undersampling.
You can also consider using a deep learning approach like LSTM which are suitable for speech and sequence data. You can refer to the following documentations for the same:
