You don't understand me!": Comparing asr results for l1 and l2 speakers of swedish

Ronald Cumbal; Birger Moell; Jośe Lopes; Olov Engwall

doi:10.21437/Interspeech.2021-2140

You don't understand me!": Comparing asr results for l1 and l2 speakers of swedish

Ronald Cumbal, Birger Moell, Jośe Lopes, Olov Engwall

School of Mathematical & Computer Sciences

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

8 Citations (Scopus)

Abstract

The performance of Automatic Speech Recognition (ASR) systems has constantly increased in state-of-the-art development. However, performance tends to decrease considerably in more challenging conditions (e.g., background noise, multiple speaker social conversations) and with more atypical speakers (e.g., children, non-native speakers or people with speech disorders), which signifies that general improvements do not necessarily transfer to applications that rely on ASR, e.g., educational software for younger students or language learners. In this study, we focus on the gap in performance between recognition results for native and non-native, read and spontaneous, Swedish utterances transcribed by different ASR services. We compare the recognition results using Word Error Rate and analyze the linguistic factors that may generate the observed transcription errors.

Original language	English
Title of host publication	Proceedings of INTERSPEECH 2021
Publisher	ISCA
Pages	4463-4467
Number of pages	5
ISBN (Electronic)	9781713836902
DOIs	https://doi.org/10.21437/Interspeech.2021-2140
Publication status	Published - 2021
Event	22nd Annual Conference of the International Speech Communication Association 2021 - Brno, Czech Republic Duration: 30 Aug 2021 → 3 Sept 2021

Conference

Conference	22nd Annual Conference of the International Speech Communication Association 2021
Abbreviated title	INTERSPEECH 2021
Country/Territory	Czech Republic
City	Brno
Period	30/08/21 → 3/09/21

Keywords

Automatic speech recognition
Language learning
Non-native speech

ASJC Scopus subject areas

Language and Linguistics
Human-Computer Interaction
Signal Processing
Software
Modelling and Simulation

Access to Document

10.21437/Interspeech.2021-2140

Cite this

@inproceedings{7917f24dbedc4c92bc26e682985e88bd,

title = "You don't understand me!{"}: Comparing asr results for l1 and l2 speakers of swedish",

abstract = "The performance of Automatic Speech Recognition (ASR) systems has constantly increased in state-of-the-art development. However, performance tends to decrease considerably in more challenging conditions (e.g., background noise, multiple speaker social conversations) and with more atypical speakers (e.g., children, non-native speakers or people with speech disorders), which signifies that general improvements do not necessarily transfer to applications that rely on ASR, e.g., educational software for younger students or language learners. In this study, we focus on the gap in performance between recognition results for native and non-native, read and spontaneous, Swedish utterances transcribed by different ASR services. We compare the recognition results using Word Error Rate and analyze the linguistic factors that may generate the observed transcription errors.",

keywords = "Automatic speech recognition, Language learning, Non-native speech",

author = "Ronald Cumbal and Birger Moell and Jo{\'s}e Lopes and Olov Engwall",

note = "Funding Information: This work was supported by the Swedish Research Council Grant 2016-03698 “Collaborative Robot Assisted Language Learning” and the VINNOVA grant 2019-02997 “AI teacher for more efficient language learning”. Publisher Copyright: Copyright {\textcopyright} 2021 ISCA.; 22nd Annual Conference of the International Speech Communication Association 2021, INTERSPEECH 2021 ; Conference date: 30-08-2021 Through 03-09-2021",

year = "2021",

doi = "10.21437/Interspeech.2021-2140",

language = "English",

pages = "4463--4467",

booktitle = "Proceedings of INTERSPEECH 2021",

publisher = "ISCA",

}

TY - GEN

T1 - You don't understand me!"

T2 - 22nd Annual Conference of the International Speech Communication Association 2021

AU - Cumbal, Ronald

AU - Moell, Birger

AU - Lopes, Jośe

AU - Engwall, Olov

N1 - Funding Information: This work was supported by the Swedish Research Council Grant 2016-03698 “Collaborative Robot Assisted Language Learning” and the VINNOVA grant 2019-02997 “AI teacher for more efficient language learning”. Publisher Copyright: Copyright © 2021 ISCA.

PY - 2021

Y1 - 2021

N2 - The performance of Automatic Speech Recognition (ASR) systems has constantly increased in state-of-the-art development. However, performance tends to decrease considerably in more challenging conditions (e.g., background noise, multiple speaker social conversations) and with more atypical speakers (e.g., children, non-native speakers or people with speech disorders), which signifies that general improvements do not necessarily transfer to applications that rely on ASR, e.g., educational software for younger students or language learners. In this study, we focus on the gap in performance between recognition results for native and non-native, read and spontaneous, Swedish utterances transcribed by different ASR services. We compare the recognition results using Word Error Rate and analyze the linguistic factors that may generate the observed transcription errors.

AB - The performance of Automatic Speech Recognition (ASR) systems has constantly increased in state-of-the-art development. However, performance tends to decrease considerably in more challenging conditions (e.g., background noise, multiple speaker social conversations) and with more atypical speakers (e.g., children, non-native speakers or people with speech disorders), which signifies that general improvements do not necessarily transfer to applications that rely on ASR, e.g., educational software for younger students or language learners. In this study, we focus on the gap in performance between recognition results for native and non-native, read and spontaneous, Swedish utterances transcribed by different ASR services. We compare the recognition results using Word Error Rate and analyze the linguistic factors that may generate the observed transcription errors.

KW - Automatic speech recognition

KW - Language learning

KW - Non-native speech

UR - http://www.scopus.com/inward/record.url?scp=85119499427&partnerID=8YFLogxK

U2 - 10.21437/Interspeech.2021-2140

DO - 10.21437/Interspeech.2021-2140

M3 - Conference contribution

AN - SCOPUS:85119499427

SP - 4463

EP - 4467

BT - Proceedings of INTERSPEECH 2021

PB - ISCA

Y2 - 30 August 2021 through 3 September 2021

ER -

You don't understand me!": Comparing asr results for l1 and l2 speakers of swedish

Abstract

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this