High Performance Segmentation of Spontaneous Speech Using Part of Speech and Trigger Word Information

41533

Knowledge Triples