A system for automatic alignment of broadcast media captions using weighted finite-state transducers

Peter Bell; Steve Renals

doi:10.1109/ASRU.2015.7404861

A system for automatic alignment of broadcast media captions using weighted finite-state transducers

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We describe our system for alignment of broadcast media captions in the 2015 MGB Challenge. A precise time alignment of previously-generated subtitles to media data is important in the process of caption generation by broadcasters. However,this task is challenging due to the highly diverse, often noisy content of the audio, and because the subtitles are frequently not a verbatim representation of the actual words spoken. Our system employs a two-pass approach with appropriately constrained weighted finite state transducers (WFSTs) to enable good alignment even when the audio quality would be challenging for conventional ASR. The system achieves an f-score of 0.8965 on the MGB Challenge development set.

Original language	English
Title of host publication	2015 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2015)
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	675-680
Number of pages	6
ISBN (Print)	978-1-4799-7291-3
DOIs	https://doi.org/10.1109/ASRU.2015.7404861
Publication status	Published - 2015

Access to Document

10.1109/ASRU.2015.7404861

master_final-1Accepted author manuscript, 265 KB

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=7404861

Natural Speech Technology
Renals, S. & King, S.
EPSRC
1/05/11 → 31/07/16
Project: Research

Cite this

@inproceedings{48815cb43057415cb3a55063f3551e3e,

title = "A system for automatic alignment of broadcast media captions using weighted finite-state transducers",

abstract = "We describe our system for alignment of broadcast media captions in the 2015 MGB Challenge. A precise time alignment of previously-generated subtitles to media data is important in the process of caption generation by broadcasters. However,this task is challenging due to the highly diverse, often noisy content of the audio, and because the subtitles are frequently not a verbatim representation of the actual words spoken. Our system employs a two-pass approach with appropriately constrained weighted finite state transducers (WFSTs) to enable good alignment even when the audio quality would be challenging for conventional ASR. The system achieves an f-score of 0.8965 on the MGB Challenge development set.",

author = "Peter Bell and Steve Renals",

year = "2015",

doi = "10.1109/ASRU.2015.7404861",

language = "English",

isbn = "978-1-4799-7291-3",

pages = "675--680",

booktitle = "2015 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2015)",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

}

A system for automatic alignment of broadcast media captions using weighted finite-state transducers. / Bell, Peter ; Renals, Steve.

2015 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2015). Institute of Electrical and Electronics Engineers (IEEE), 2015. p. 675-680.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A system for automatic alignment of broadcast media captions using weighted finite-state transducers

AU - Bell, Peter

AU - Renals, Steve

PY - 2015

Y1 - 2015

N2 - We describe our system for alignment of broadcast media captions in the 2015 MGB Challenge. A precise time alignment of previously-generated subtitles to media data is important in the process of caption generation by broadcasters. However,this task is challenging due to the highly diverse, often noisy content of the audio, and because the subtitles are frequently not a verbatim representation of the actual words spoken. Our system employs a two-pass approach with appropriately constrained weighted finite state transducers (WFSTs) to enable good alignment even when the audio quality would be challenging for conventional ASR. The system achieves an f-score of 0.8965 on the MGB Challenge development set.

AB - We describe our system for alignment of broadcast media captions in the 2015 MGB Challenge. A precise time alignment of previously-generated subtitles to media data is important in the process of caption generation by broadcasters. However,this task is challenging due to the highly diverse, often noisy content of the audio, and because the subtitles are frequently not a verbatim representation of the actual words spoken. Our system employs a two-pass approach with appropriately constrained weighted finite state transducers (WFSTs) to enable good alignment even when the audio quality would be challenging for conventional ASR. The system achieves an f-score of 0.8965 on the MGB Challenge development set.

U2 - 10.1109/ASRU.2015.7404861

DO - 10.1109/ASRU.2015.7404861

M3 - Conference contribution

SN - 978-1-4799-7291-3

SP - 675

EP - 680

BT - 2015 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2015)

PB - Institute of Electrical and Electronics Engineers (IEEE)

ER -

A system for automatic alignment of broadcast media captions using weighted finite-state transducers

Abstract

Access to Document

Fingerprint

Projects

Natural Speech Technology

Cite this