Comment by notduncansmith

The repository (https://github.com/facebookresearch/wav2letter) claims to come with pre-trained models for automated speech recognition.