Audio-Augment

Audio data augmentation tool for machine learning projects

This library augments audio training datasets by transforming provided WAV files in 10 different ways:

1. Waveform Inversion
2. Highpass Filter
3. Lowpass Filter
4. Bandpass Filter
5. Add noise (normal, uniform)
6. Pitch shift (low, high)
7. Time shift (slow, fast)

Directory Setup

Create two folders in your current working directory, one called unprocessed and another called processed. Place all audio samples that you want to transform in the unprocessed folder (must be WAV format). Run this method and it will transform all audio files in your unprocessed folder, and save the new set of augmented samples in the processed folder.

Requirements

• numpy

• pandas

• matplotlib

• soundfile

• librosa

    For questions about this library, email wesleylaurencetech@gmail.com