SpeakerLDA: Discovering Topics in Transcribed Multi-Speaker Audio Contents

  • Damiano Spina
  • Johanne R. Trippas
  • Lawrence Cavedon
  • Mark Sanderson
Proceedings of ACM Multimedia 2015 Workshop on Speech, Language and Audio in Multimedia (SLAM'15), 2015

Topic models such as Latent Dirichlet Allocation (LDA) have been extensively used for characterizing text collections according to the topics discussed in documents. Organizing documents according to topic can be applied to different information access tasks such as document clustering, content-based recommendation or summarization. Spoken documents such as podcasts typically involve more than one speaker (e.g., meetings, interviews, chat shows or news with reporters). This paper presents a work-in-progress based on a variation of LDA that includes in the model the different speakers participating in conversational audio transcripts. Intuitively, each speaker has her own background knowledge which generates different topic and word distributions. We believe that informing a topic model with speaker segmentation (e.g., using existing speaker diarization techniques) may enhance discovery of topics in multi-speaker audio content.

@InProceedings{spina2015speakerlda,
author={Spina, Damiano and Trippas, Johanne R. and  
                 Cavedon, Lawrence and Sanderson, Mark},
title={SpeakerLDA: Discovering Topics in Transcribed Multi-Speaker Audio Contents},
booktitle={Proceedings of ACM Multimedia 2015 Workshop on 
                    Speech, Language and Audio in Multimedia (SLAM'15)},
year={2015}
}

thumbnail

Presentation

Damiano Spina
Assign a menu in the Right Menu options.