Large-scale corpus of spontaneous spoken Japanese containing over 700 hours of speech data, essential for speech recognition and prosody research.