Introduction

• This repo is made for updating GSoC 2024 project proposal with some helper codes that I for the project proposed.
• Please download relevant data of audioset from the link
• Download samples of People's Speech Dataset from here link
• Note that you can also use utils.ipynb as a scrip to download these datasets.
• Other tools that I have tried for the proposal include - yt-dlp, shot_detection, video_llava

Videos with these Labels and childs of these to be avoided

/m/0l14jd: Choir /m/0kpv1t: Music genre /m/074ft: Song

/m/09x0r: Speech /m/02rtxlg: Whispering /m/015lz1: Singing /t/dd00089: Miscellaneous sources

/m/09l8g: Human voice -> Note it is better to add it later for some of them which are not speech.