elmargb / speaker-recognition Goto Github PK

#Speaker-Recognition using Gaussian Mixture Models

##Overview

Final project for the course ECE446 - Sensory Communication at University of Toronto. The project consists in a Speaker Recognition system that uses Gaussian Mixture Models (GMM) and a report that explains the entire system can be found here.

Dependencies

The system has dependencies with the following libraries:

• Scipy
• Numpy
• Scikit-Learn

Database

First it is needed to build a database of users. Voice samples of each user in the database are recorded and saved as .wav files at ./Database/<username>/, where <username> is the name of each user. In this case, more samples means more accuracy.

The samples are text-independent, i.e. the user can say anything and the system will still work.

How to run

The file that need to be run is the extract.py. It will uses the file at ./Test/ as the one to be recognized. As the program start running, appropriate outputs appear showing the results.

Observations

The code is not well organized, and it needs to be improved. This probably will be solved in the future, when I have free time.