A free audio dataset of spoken digits. Think MNIST for audio. - View it on GitHub