A deep neural network approach for sentence boundary detection in broadcast news

Xu, Chenglin; Xie, Lei; Huang, Gongsheng; Xiao, Xiong; Chng, Eng Siong; Li, Haizhou

doi:10.21437/interspeech.2014-599

Cited by 26 publications

(7 citation statements)

References 18 publications

(19 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Early studies on the punctuation restoration task explored a wide range of features such as lexical, acoustic, prosodic, and their combination (Gravano et al, 2009;Levy et al, 2012;Xu et al, 2014;Che et al, 2016a;Szaszák and Tündik, 2019). Graphical model such as conditional random field has been widely used for this task (Lu and Ng, 2010;Zhang et al, 2013) before the emerging of neural network.…”

Section: Related Workmentioning

confidence: 99%

SemEval 2022 Task 12: Symlink- Linking Mathematical Symbols to their Descriptions

Lai¹,

Veyseh²,

Dernoncourt³

et al. 2022

Preprint

View full text Add to dashboard Cite

Given the increasing number of livestreaming videos, automatic speech recognition and post-processing for livestreaming video transcripts are crucial for efficient data management as well as knowledge mining. A key step in this process is punctuation restoration which restores fundamental text structures such as phrase and sentence boundaries from the video transcripts. This work presents a new human-annotated corpus, called BehancePR, for punctuation restoration in livestreaming video transcripts. Our experiments on BehancePR demonstrate the challenges of punctuation restoration for this domain. Furthermore, we show that popular natural language processing toolkits are incapable of detecting sentence boundary on non-punctuated transcripts of livestreaming videos, calling for more research effort to develop robust models for this area.

show abstract

Section: Related Workmentioning

confidence: 99%

SemEval 2022 Task 12: Symlink- Linking Mathematical Symbols to their Descriptions

Lai¹,

Veyseh²,

Dernoncourt³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Cuendet et al (2006) makes use of a range of lexical and acoustic features. Xu et al (2014) used prosodic and lexical features to implement a DNN combined with a CRF classifier for broadcast news speech. Atterer et al (2008) used syntactic ground-truth information, in a rare incremental approach, to predict whether the current word on Switchboard is the end of the utterance (dialog act).…”

Section: Related Workmentioning

confidence: 99%

Re-framing Incremental Deep Language Models for Dialogue Processing with Multi-task Learning

Rohanian¹,

Hough²

2020

Preprint

View full text Add to dashboard Cite

We present a multi-task learning framework to enable the training of one universal incremental dialogue processing model with four tasks of disfluency detection, language modelling, part-ofspeech tagging, and utterance segmentation in a simple deep recurrent setting. We show that these tasks provide positive inductive biases to each other with the optimal contribution of each one relying on the severity of the noise from the task. Our live multi-task model outperforms similar individual tasks, delivers competitive performance, and is beneficial for future use in conversational agents in psychiatric treatment.

show abstract

“…They rely on the transcript content to extract features like bag-of-word, POS tags or word embeddings [7,12,16,18,24,27,31]. Mixture of acoustic and lexical features have also been explored [1,13,14,33], which is advantageous when both audio signal and transcript are available.…”

Section: Sentence Boundary Evaluationmentioning

confidence: 99%

WiSeBE: Window-Based Sentence Boundary Evaluation

González-Gallardo

Torres‐Moreno

2018

Advances in Computational Intelligence

View full text Add to dashboard Cite

Sentence Boundary Detection (SBD) has been a major research topic since Automatic Speech Recognition transcripts have been used for further Natural Language Processing tasks like Part of Speech Tagging, Question Answering or Automatic Summarization. But what about evaluation? Do standard evaluation metrics like precision, recall, F-score or classification error; and more important, evaluating an automatic system against a unique reference is enough to conclude how well a SBD system is performing given the final application of the transcript? In this paper we propose Window-based Sentence Boundary Evaluation (WiSeBE), a semi-supervised metric for evaluating Sentence Boundary Detection systems based on multi-reference (dis)agreement. We evaluate and compare the performance of different SBD systems over a set of Youtube transcripts using WiSeBE and standard metrics. This double evaluation gives an understanding of how WiSeBE is a more reliable metric for the SBD task.

show abstract

A deep neural network approach for sentence boundary detection in broadcast news

Cited by 26 publications

References 18 publications

SemEval 2022 Task 12: Symlink- Linking Mathematical Symbols to their Descriptions

SemEval 2022 Task 12: Symlink- Linking Mathematical Symbols to their Descriptions

Re-framing Incremental Deep Language Models for Dialogue Processing with Multi-task Learning

WiSeBE: Window-Based Sentence Boundary Evaluation

Contact Info

Product

Resources

About