This repository contains the Learning from Unstructured Data project developed during the Erasmus program at Nova FCT in Lisbon in Big Data Analytics and Engineering.
The aim of the project is to use different types of deep neural networks to solve a multi-class classification problem on audio data (wav file). The implementation of the networks are done using in Tensorflow/Keras.
Three types of deep models are employed to solve the task:
- 1D Convolution Network and 2D Convolution Network
- Recurrent Network
- Transformer Network
Each model is trained and evaluated on both raw data (wav files) and on the Short Time Fourier Transformation (STFT).
In this repository, in the root level, you can find the pdf report of the project and the jupyter notebook where the models are implemented.