This project is the application of audio file processing using the concept of MFCCs(Mel-Frequency Cepstrum Coeffs) executed using the librosa library in combination with machine learning concept (Logisitic Regression) to create a ML model that can predict the covid status. Our model takes audio file of any user as input, evaluates then outputs the prediction.
The features of this project are :
- MFCCs
- k-class Logisitic Regression
- 94% accuracy (with binary Classification with non-uniform training)
- programing language - python
As of now only local deployment is possible
The python code file and application, open and download download here
Install Prerequisite
The python code file of the model info, open and download download here
You will require the following versions of the mentioned libraries in the file requirements.txt(linked below) to implement this as is,
suggestion - download from here
💭 currently in work:
👉 Modified application of training data to generate a better learning model by readjusting training data to be uniform.
👉 Improving mulitclass classification accuracy current accuracy ≃ 80%
👉 Application on wider range of test data set.
B.E.Pranav Kumaar
Student ID @Amrita Vishwa Vidyapeetham - CB.EN.U4AIE20052❄️ Github
Divi Eswar Choudary
Student ID @Amrita Vishwa Vidyapeetham - CB.EN.U4AIE20012❄️ Github