Attention-Based End-To-End Named Entity Recognition From Speech

Dejan Porjazovski; Juho Leinonen; Mikko Kurimo

doi:10.1007/978-3-030-83527-9_40

Attention-Based End-To-End Named Entity Recognition From Speech

Dejan Porjazovski, Juho Leinonen, Mikko Kurimo

Speech Recognition

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

4 Citations (Scopus)

213 Downloads (Pure)

Abstract

Named entities are heavily used in the field of spoken language understanding, which uses speech as an input. The standard way of doing named entity recognition from speech involves a pipeline of two systems, where first the automatic speech recognition system generates the transcripts, and then the named entity recognition system produces the named entity tags from the transcripts. In such cases, automatic speech recognition and named entity recognition systems are trained independently, resulting in the automatic speech recognition branch not being optimized for named entity recognition and vice versa. In this paper, we propose two attention-based approaches for extracting named entities from speech in an end-to-end manner, that show promising results. We compare both attention-based approaches on Finnish, Swedish, and English data sets, underlining their strengths and weaknesses.

Original language	English
Title of host publication	Text, Speech, and Dialogue - 24th International Conference, TSD 2021, Proceedings
Editors	Kamil Ekštein, František Pártl, Miloslav Konopík
Publisher	Springer
Pages	469 - 480
Number of pages	12
ISBN (Electronic)	978-3-030-83527-9
ISBN (Print)	9783030835262
DOIs	https://doi.org/10.1007/978-3-030-83527-9_40
Publication status	Published - 2021
MoE publication type	A4 Conference publication
Event	International Conference on Text, Speech, and Dialogue - Olomouc, Czech Republic Duration: 6 Sept 2021 → 9 Sept 2021 Conference number: 24

Publication series

Name	Lecture Notes in Computer Science
Volume	12848
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Conference on Text, Speech, and Dialogue
Abbreviated title	TSD
Country/Territory	Czech Republic
City	Olomouc
Period	06/09/2021 → 09/09/2021

Access to Document

10.1007/978-3-030-83527-9_40

TSD_2021_Attention_Based_End_To_End_Named_Entity_Recognition_From_SpeechAccepted author manuscript, 1.15 MB

Cite this

@inproceedings{d4014ded6a724a81bad8beeea6ad2e2b,

title = "Attention-Based End-To-End Named Entity Recognition From Speech",

abstract = "Named entities are heavily used in the field of spoken language understanding, which uses speech as an input. The standard way of doing named entity recognition from speech involves a pipeline of two systems, where first the automatic speech recognition system generates the transcripts, and then the named entity recognition system produces the named entity tags from the transcripts. In such cases, automatic speech recognition and named entity recognition systems are trained independently, resulting in the automatic speech recognition branch not being optimized for named entity recognition and vice versa. In this paper, we propose two attention-based approaches for extracting named entities from speech in an end-to-end manner, that show promising results. We compare both attention-based approaches on Finnish, Swedish, and English data sets, underlining their strengths and weaknesses. ",

author = "Dejan Porjazovski and Juho Leinonen and Mikko Kurimo",

note = "| openaire: EC/H2020/780069/EU//MeMAD ; International Conference on Text, Speech, and Dialogue, TSD ; Conference date: 06-09-2021 Through 09-09-2021",

year = "2021",

doi = "10.1007/978-3-030-83527-9_40",

language = "English",

isbn = "9783030835262",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "469 -- 480",

editor = "Kamil Ek{\v s}tein and Franti{\v s}ek P{\'a}rtl and Miloslav Konop{\'i}k",

booktitle = "Text, Speech, and Dialogue - 24th International Conference, TSD 2021, Proceedings",

address = "Germany",

}

Porjazovski, D , Leinonen, J & Kurimo, M 2021, Attention-Based End-To-End Named Entity Recognition From Speech. in K Ekštein, F Pártl & M Konopík (eds), Text, Speech, and Dialogue - 24th International Conference, TSD 2021, Proceedings. Lecture Notes in Computer Science, vol. 12848 , Springer, pp. 469 - 480, International Conference on Text, Speech, and Dialogue, Olomouc, Czech Republic, 06/09/2021. https://doi.org/10.1007/978-3-030-83527-9_40

Attention-Based End-To-End Named Entity Recognition From Speech. / Porjazovski, Dejan ; Leinonen, Juho ; Kurimo, Mikko.
Text, Speech, and Dialogue - 24th International Conference, TSD 2021, Proceedings. ed. / Kamil Ekštein; František Pártl; Miloslav Konopík. Springer, 2021. p. 469 - 480 (Lecture Notes in Computer Science; Vol. 12848 ).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Attention-Based End-To-End Named Entity Recognition From Speech

AU - Porjazovski, Dejan

AU - Leinonen, Juho

AU - Kurimo, Mikko

N1 - Conference code: 24

PY - 2021

Y1 - 2021

N2 - Named entities are heavily used in the field of spoken language understanding, which uses speech as an input. The standard way of doing named entity recognition from speech involves a pipeline of two systems, where first the automatic speech recognition system generates the transcripts, and then the named entity recognition system produces the named entity tags from the transcripts. In such cases, automatic speech recognition and named entity recognition systems are trained independently, resulting in the automatic speech recognition branch not being optimized for named entity recognition and vice versa. In this paper, we propose two attention-based approaches for extracting named entities from speech in an end-to-end manner, that show promising results. We compare both attention-based approaches on Finnish, Swedish, and English data sets, underlining their strengths and weaknesses.

AB - Named entities are heavily used in the field of spoken language understanding, which uses speech as an input. The standard way of doing named entity recognition from speech involves a pipeline of two systems, where first the automatic speech recognition system generates the transcripts, and then the named entity recognition system produces the named entity tags from the transcripts. In such cases, automatic speech recognition and named entity recognition systems are trained independently, resulting in the automatic speech recognition branch not being optimized for named entity recognition and vice versa. In this paper, we propose two attention-based approaches for extracting named entities from speech in an end-to-end manner, that show promising results. We compare both attention-based approaches on Finnish, Swedish, and English data sets, underlining their strengths and weaknesses.

UR - http://www.scopus.com/inward/record.url?scp=85115247889&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-83527-9_40

DO - 10.1007/978-3-030-83527-9_40

M3 - Conference article in proceedings

SN - 9783030835262

T3 - Lecture Notes in Computer Science

SP - 469

EP - 480

BT - Text, Speech, and Dialogue - 24th International Conference, TSD 2021, Proceedings

A2 - Ekštein, Kamil

A2 - Pártl, František

A2 - Konopík, Miloslav

PB - Springer

T2 - International Conference on Text, Speech, and Dialogue

Y2 - 6 September 2021 through 9 September 2021

ER -

Attention-Based End-To-End Named Entity Recognition From Speech

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

MeMAD: Methods for Managing Audiovisual Data: Combining Automatic Efficiency with Human Accuracy

Science-IT

Cite this

Attention-Based End-To-End Named Entity Recognition From Speech

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Projects

MeMAD: Methods for Managing Audiovisual Data: Combining Automatic Efficiency with Human Accuracy

Equipment

Science-IT

Cite this