MaisyZhang / spec_augment

🔦 A Pytorch implementation of GoogleBrain's SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition

SpecAugment.py

A Pytorch implementation of GoogleBrain's SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition

SpecAugment is a SOTA-achieving data augmentation approach on speech recognition. The paper's authors did not publish code that I could find and their implementation was in TensorFlow.

To use:

1. run install.sh (I recommend to use a unique conda env for the project)
2. Check out SpecAugment.ipynb (a Jupyter notebook) for the functions.

Augmentations

1. Time Warp (DONE!)

2. Time Mask (DONE!)

3. Frequency Mask (DONE!)

Let's be friends!

• @zachcaceres
• zach.dev