Birds Audio Classification using CNN

Project Goal:

The goal of this project is to develop a convolutional neural network (CNN) model that can classify bird species from their audio recordings.

Project Background:

Bird sound classification is a challenging task due to the high variability in bird calls, the presence of background noise, and the need to classify a large number of species. However, CNNs have been shown to be effective in image classification tasks, and they have also been used for audio classification tasks.

Project Methodology:

• The proposed methodology will involve the following steps:
• Collect a dataset of audio recordings of bird calls.
• Extract features from the audio recordings, such as the spectrogram.
• Train a CNN model on the extracted features.
• Evaluate the performance of the CNN model on a test set of audio recordings.

Prerequiste

• Python(v3.10.2)
• Modules used:
  • Tensorflow (v2.10.0)
  • pandas (v1.4.1)
  • scikit-learn (v1.1.2)
  • numpy (v1.24.4)
  • librosa (v0.10.0.post2) => For audio processing
  • opencv-python (v4.6.0.66)
• IDE used : VSCode

Usage

1. Clone the Repository

git clone https://github.com/SonicWarrior1/Birds-Audio-Classification.git

2. First run the data_preperation.ipynb to download the audio set and generate spectrograms for the audios.

3. Now run the model_train.ipynb file to train the CNN and save the model.

4. Run the Flask app file app.py using the following command in your terminal.

flask run

5. Now select the audio file for input and press the submit button to predict the bird class.