A large collection of audio recordings and transcripts used for developing and testing speech recognition systems.