Motivation:
This project is a term project I am working under my course "Statistical Foundation for AI/ML" as a part of my curriculum. However, I have an idea which will extend my content of my project and I want to attempt to explore it in a sense that the model I attempt to create for my project to be totally unsupervised.
Topic:
I originally attempt to explore and compare the accuracies of different types of Hidden Markov Models for speaker diarisation. The basic disadvantage of any Hidden Markov Model is that it requires the number of components as a pre-requisite for the model to run. I, in my extended project, attempt to create an "adaptive" hidden markov model so that I can make the model completely unsupervised.
Edit:
I have done this project, and is now readily available on my Github repo here.
No comments:
Post a Comment