Fix a typo in sample rate. (#2)

Added medium and low models

Added voices.json and MODEL_CARD. Fixed location of glados-turret -> glados-turret/high

Moved models to subfolders similar to original piper-voices repo