Audio Classification Features

It is made to extract the features from any audio dataset. User's have to provide location of the dataset folder and this library will produce x and y npy files. We also provide custom built Keras model for training.

Installation

$ pip install audio_classification_features

Usage

Making Training Dataset

from audio_classification_features import audio_features as af
af().extractor('dataset')

These npy files are loaded with numpy using following commands:

import numpy
x=np.load('x.npy')
y=np.load('y.npy')

Training Custom Built Model

#input_shape generated by this package is of shape (9,13,1)
input_shape=(9,13,1)

Making Predictions

from audio_classification_features import audio_features as af

#filename example: audio_test.wav
#num_classes same as above 
#model

af().make_prediction(filename,num_classes,model)

Dataset Structure

folder name of audio will be used as label.

• dataset
  • audio class 1
    • audiofile.wav
  • audio class 2
    • audiofile.wav

LICENSE

GNU General Public License v3.0