/audio_classification

Content-based Audio Classification

Primary LanguagePython

Content-based Audio Classification

Plug and play audio classification. It scored 96% for snoring detection as benchmark.
You can try your own audio dataset.
This repository support different types of feature extraction methods and classifiers.

Expert Model

Currently, k-NN and Logistic Regression are available

DNN Model

Currently, MLP, CNN, ResNet(Transfer Learning) are supported

How to install

Make sure to use python 3.6
pip install -r backend/requirements.txt

Prepare data

  1. Make folder called data with sub-folder indicates categories.
  2. Move audio files into each sub-folders

How to run

Now, you are good to go Simply, run python backend/src/main.py

Parameter setting, feature extraction methods or different classifiers

Edit config/master_config.ini
You can select different types of feature extraction and classifier.
I recommend to set audio length to median of audio files.