\n
@techreport{\n title = {A CRF-based Approach to Automatic Disfluency Detection in a French Call-Centre Corpus},\n type = {techreport},\n year = {2014},\n keywords = {Index Terms : disfluencies,conditional random fields,conver-sational speech,spontaneous speech},\n websites = {https://gforge.inria.fr/projects/discretize4crf/.},\n id = {496020df-f2c6-3537-b35e-f521e49636dc},\n created = {2019-05-16T08:21:33.452Z},\n accessed = {2019-05-16},\n file_attached = {true},\n profile_id = {1bff199d-3fb6-39f8-9a95-8103f3a5d433},\n group_id = {af6234f1-5a5f-3882-bb01-83ebf2615cfd},\n last_modified = {2019-05-16T08:21:35.038Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {false},\n hidden = {false},\n private_publication = {false},\n abstract = {In this paper, we present a Conditional Random Field based approach for automatic detection of edit disfluencies in a conversational telephone corpus in French. We define dis-fluency patterns using both linguistic and acoustic features to perform disfluency detection. Two related tasks are considered : the first task aims at detecting the disfluent speech portion proper or reparandum, i.e. the portion to be removed if we want to improve the readability of transcribed data ; in the second task, we aim at identifying also the corrected portion or repair which can be useful in follow-up discourse and dialogue analyses or in opinion mining. For these two tasks, we present comparative results as a function of the involved type of features (acoustic and/or linguistic). Generally speaking, best results are obtained by CRF models combining both acoustic and linguistic features.},\n bibtype = {techreport},\n author = {Dutrey, Camille and Clavel, Chloé and Rosset, Sophie and Vasilescu, Ioana and Adda-Decker, Martine}\n}
\n
\n In this paper, we present a Conditional Random Field based approach for automatic detection of edit disfluencies in a conversational telephone corpus in French. We define dis-fluency patterns using both linguistic and acoustic features to perform disfluency detection. Two related tasks are considered : the first task aims at detecting the disfluent speech portion proper or reparandum, i.e. the portion to be removed if we want to improve the readability of transcribed data ; in the second task, we aim at identifying also the corrected portion or repair which can be useful in follow-up discourse and dialogue analyses or in opinion mining. For these two tasks, we present comparative results as a function of the involved type of features (acoustic and/or linguistic). Generally speaking, best results are obtained by CRF models combining both acoustic and linguistic features.\n