ISCA Archive Interspeech 2021
ISCA Archive Interspeech 2021

Auxiliary Sequence Labeling Tasks for Disfluency Detection

Dongyub Lee, Byeongil Ko, Myeong Cheol Shin, Taesun Whang, Daniel Lee, Eunhwa Kim, Eunggyun Kim, Jaechoon Jo

Detecting disfluencies in spontaneous speech is an important preprocessing step in natural language processing and speech recognition applications. Existing works for disfluency detection have focused on designing a single objective only for disfluency detection, while auxiliary objectives utilizing linguistic information of a word such as named entity or part-of-speech information can be effective. In this paper, we focus on detecting disfluencies on spoken transcripts and propose a method utilizing named entity recognition (NER) and part-of-speech (POS) as auxiliary sequence labeling (SL) tasks for disfluency detection. First, we investigate cases that utilizing linguistic information of a word can prevent mispredicting important words and can be helpful for the correct detection of disfluencies. Second, we show that training a disfluency detection model with auxiliary SL tasks can improve its F-score in disfluency detection. Then, we analyze which auxiliary SL tasks are influential depending on baseline models. Experimental results on the widely used English Switchboard dataset show that our method outperforms the previous state-of-the-art in disfluency detection.

doi: 10.21437/Interspeech.2021-400

Cite as: Lee, D., Ko, B., Shin, M.C., Whang, T., Lee, D., Kim, E., Kim, E., Jo, J. (2021) Auxiliary Sequence Labeling Tasks for Disfluency Detection. Proc. Interspeech 2021, 4229-4233, doi: 10.21437/Interspeech.2021-400

  author={Dongyub Lee and Byeongil Ko and Myeong Cheol Shin and Taesun Whang and Daniel Lee and Eunhwa Kim and Eunggyun Kim and Jaechoon Jo},
  title={{Auxiliary Sequence Labeling Tasks for Disfluency Detection}},
  booktitle={Proc. Interspeech 2021},