Jovana-Gentic / Bird_song_classification

Multiclass audio classification of bird calls.

Bird song classification

Jovana Gentić 🦆

kaggle notebook

"Bird song classification" is an audio classification model. We used 1D Convolutional Neural Networks (CNN) to learn a multiclass classifier on dataset of 5422 bird songs, belonging to 5 classes. We used spectrograms as inputs to the model . This repo is written in Tensorflow.

Installation

To download dataset, go to https://www.kaggle.com/datasets/vinayshanbhag/bird-song-data-set/data. Dataset folder should be called dataset and it should be placed in the same folder as data.py.

pip install -r requirements

Model training

Running the code

To train the model, execute command:

python train.py

Model Evaluation

To evaluate the model, execute command:

python evaluate.py