Final project

By Kaare and Tobias

Resources:

Kaggle:

Official
https://www.kaggle.com/c/birdclef-2021/overview
https://www.kaggle.com/stefankahl/birdclef2021-exploring-the-data
https://www.kaggle.com/stefankahl/birdclef2021-processing-audio-data
https://www.kaggle.com/stefankahl/birdclef2021-model-training

Extra
https://www.kaggle.com/shreyasajal/birdclef-2021-librosa-audio-feature-extraction
https://www.kaggle.com/shreyasajal/audio-albumentations-torchaudio-audiomentations

YouTube:

Audio Classification with Machine Learning (EuroPython 2019) - https://www.youtube.com/watch?v=uCGROOUO_wY (https://github.com/jonnor/ESC-CNN-microcontroller)
Who's singing? Automatic bird sound recognition with machine learning - Dan Stowell - https://www.youtube.com/watch?v=pzmdOETnhI0
How convolutional neural networks work, in depth - https://www.youtube.com/watch?v=JB8T_zN7ZC0

Libraries

openl3 - https://github.com/marl/openl3

Background

Bird vocalization - https://en.wikipedia.org/wiki/Bird_vocalization

Notes:

Quality of labeling

Data Augmentation:

Audio
- Timeshifting
- Timestretch
- Pitch
- Noise
  - add (wablr open data ??)
  - remove
Extra
- GPS ?

Ethics

Who may ideally benefit from the technology? Could it do potential harm?