ISCA Archive Interspeech 2013
ISCA Archive Interspeech 2013

Intensive acoustic models constructed by integrating low-occurrence models for spoken term detection

Shiro Narumi, Kazuma Konno, Takuya Nakano, Yoshiaki Itoh, Kazunori Kojima, Masaaki Ishigame, Kazuyo Tanaka, Shi-wook Lee

Triphone acoustic models are often used as subword models for detecting out-of-vocabulary query terms in Spoken Term Detection (STD) systems. Our preliminary experiments revealed that the training data for a large portion of the approximately 8,000 triphone models are insufficient. Assuming that such insufficient models deteriorate the performance of STD, this paper proposes intensive triphone models constructed by integrating low-occurrence triphone models into high-occurrence ones. Experiments conducted using an actual lecture speech corpus showed that the proposed method improves the STD performance with regard to both triphones and demiphones, demonstrating its effectiveness.


doi: 10.21437/Interspeech.2013-6

Cite as: Narumi, S., Konno, K., Nakano, T., Itoh, Y., Kojima, K., Ishigame, M., Tanaka, K., Lee, S.-w. (2013) Intensive acoustic models constructed by integrating low-occurrence models for spoken term detection. Proc. Interspeech 2013, 25-28, doi: 10.21437/Interspeech.2013-6

@inproceedings{narumi13_interspeech,
  author={Shiro Narumi and Kazuma Konno and Takuya Nakano and Yoshiaki Itoh and Kazunori Kojima and Masaaki Ishigame and Kazuyo Tanaka and Shi-wook Lee},
  title={{Intensive acoustic models constructed by integrating low-occurrence models for spoken term detection}},
  year=2013,
  booktitle={Proc. Interspeech 2013},
  pages={25--28},
  doi={10.21437/Interspeech.2013-6}
}