This repository contains all the material used for development of Blind Source Separation based concurrent speaker localization algorithm. The artifacts are organized as follows:
This folder contains all the recording files used in the development and testing of the algorithm. The audio files are named so as to represent the real location of the speakers on ground in degrees. The naming convention used is as below:
<loc_speaker_one>_<loc_speaker_two>[_<loc_speaker_three>]
There are two types of recording w.r.t the number of concurrent speakers: Two speaker and three speaker. The files are grouped under separate folders.
Contains user manuals and important URLs related to the Seeed Studio Respeaker Core v2.
Contains a PDF presentation of the work down under concurrent speaker localization.
Will contain the source code of the developed algorithm. The source code is currently being cleaned and properly formatted for better understanding and readability. Will be publicly available soon.
Contains my MSc thesis in PDF form.