This project is based on speech analysis technology, machine learning, and deep learning algorithms and its applications. The project code has tested and trained the data according to the ranges of the features to detect the extent of hypertension. So, if we input an audio clip, using the project model we can detect if the person is hyper tensed or not.
This project specifically deals with males in the range of age 25 to 40yrs.
The dataset has been collected from platforms including Netflix and YouTube.
Voice has certain features like jitter, shimmer, pitch, etc. we can study these values and then with the help of which we can determine the results. First part is the feature extraction.
After analyzing these features, we'd predict if the person is hyper tensed or not. We have used the surfboard and librosa library in python to directly extract the audio features which are required for classification. Different algorithms like SVM, logistic regression are used.
Current Accuracy: 70%