Encouraging Internal Representations with Speaker Information in End-to-End Neural Diarization by Adding Speaker Loss | Publicación