Lessons from Building Acoustic Models with a Million Hours of Speech

Open in new window