Automatic Recognition of Target Words in Infant-Directed Speech

A. van der Klis; F.W. Adriaans; M. Han; R.W.J. Kager

doi:10.1145/3395035.3425184

Automatic Recognition of Target Words in Infant-Directed Speech

A. van der Klis, F.W. Adriaans, M. Han, R.W.J. Kager

ILS LAPD

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

This study assesses the performance of a state-of-the-art automatic speech recognition (ASR) system at extracting target words in two different speech registers: infant-directed speech (IDS) and adult-directed speech (ADS). We used the Kaldi-NL ASR-service, developed by the Dutch Foundation of Open Speech Technology. The results indicate that the accuracy of the tool is much lower in IDS than in ADS. There are differences between IDS and ADS which negatively affect the performance of the existing ASR system. Therefore, new tools need to be developed for the automatic annotation of IDS. Nevertheless, the ASR system can already find more than half of the target words, which is promising.

Original language	English
Title of host publication	ICMI '20 Companion
Subtitle of host publication	Companion Publication of the 2020 International Conference on Multimodal Interaction
Place of Publication	New York
Publisher	Association for Computing Machinery
Pages	522
Number of pages	1
ISBN (Print)	978-1-4503-8002-7
DOIs	https://doi.org/10.1145/3395035.3425184
Publication status	Published - 2020
Event	ACM International Conference on Multimodal Interaction - ICMI2020 is going virtual, Utrecht, Netherlands Duration: 25 Oct 2020 → 29 Oct 2020 Conference number: 22

Conference

Conference	ACM International Conference on Multimodal Interaction
Country/Territory	Netherlands
City	Utrecht
Period	25/10/20 → 29/10/20

Keywords

infant-directed speech
automatic speech recognition
keyword extraction
speech registers

Access to Document

10.1145/3395035.3425184

3395035.3425184Final published version, 739 KBLicence: Taverne

Cite this

@inproceedings{f9fd5afe835643a9b14efd0b75a00961,

title = "Automatic Recognition of Target Words in Infant-Directed Speech",

abstract = "This study assesses the performance of a state-of-the-art automatic speech recognition (ASR) system at extracting target words in two different speech registers: infant-directed speech (IDS) and adult-directed speech (ADS). We used the Kaldi-NL ASR-service, developed by the Dutch Foundation of Open Speech Technology. The results indicate that the accuracy of the tool is much lower in IDS than in ADS. There are differences between IDS and ADS which negatively affect the performance of the existing ASR system. Therefore, new tools need to be developed for the automatic annotation of IDS. Nevertheless, the ASR system can already find more than half of the target words, which is promising.",

keywords = "infant-directed speech, automatic speech recognition, keyword extraction, speech registers",

author = "{van der Klis}, A. and F.W. Adriaans and M. Han and R.W.J. Kager",

year = "2020",

doi = "10.1145/3395035.3425184",

language = "English",

isbn = "978-1-4503-8002-7",

pages = "522",

booktitle = "ICMI '20 Companion",

publisher = "Association for Computing Machinery",

address = "United States",

note = "ACM International Conference on Multimodal Interaction ; Conference date: 25-10-2020 Through 29-10-2020",

}

van der Klis, A , Adriaans, FW , Han, M & Kager, RWJ 2020, Automatic Recognition of Target Words in Infant-Directed Speech. in ICMI '20 Companion: Companion Publication of the 2020 International Conference on Multimodal Interaction. Association for Computing Machinery, New York, pp. 522, ACM International Conference on Multimodal Interaction, Utrecht, Netherlands, 25/10/20. https://doi.org/10.1145/3395035.3425184

Automatic Recognition of Target Words in Infant-Directed Speech. / van der Klis, A.; Adriaans, F.W.; Han, M. et al.
ICMI '20 Companion: Companion Publication of the 2020 International Conference on Multimodal Interaction. New York: Association for Computing Machinery, 2020. p. 522.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Automatic Recognition of Target Words in Infant-Directed Speech

AU - van der Klis, A.

AU - Adriaans, F.W.

AU - Han, M.

AU - Kager, R.W.J.

N1 - Conference code: 22

PY - 2020

Y1 - 2020

N2 - This study assesses the performance of a state-of-the-art automatic speech recognition (ASR) system at extracting target words in two different speech registers: infant-directed speech (IDS) and adult-directed speech (ADS). We used the Kaldi-NL ASR-service, developed by the Dutch Foundation of Open Speech Technology. The results indicate that the accuracy of the tool is much lower in IDS than in ADS. There are differences between IDS and ADS which negatively affect the performance of the existing ASR system. Therefore, new tools need to be developed for the automatic annotation of IDS. Nevertheless, the ASR system can already find more than half of the target words, which is promising.

AB - This study assesses the performance of a state-of-the-art automatic speech recognition (ASR) system at extracting target words in two different speech registers: infant-directed speech (IDS) and adult-directed speech (ADS). We used the Kaldi-NL ASR-service, developed by the Dutch Foundation of Open Speech Technology. The results indicate that the accuracy of the tool is much lower in IDS than in ADS. There are differences between IDS and ADS which negatively affect the performance of the existing ASR system. Therefore, new tools need to be developed for the automatic annotation of IDS. Nevertheless, the ASR system can already find more than half of the target words, which is promising.

KW - infant-directed speech

KW - automatic speech recognition

KW - keyword extraction

KW - speech registers

U2 - 10.1145/3395035.3425184

DO - 10.1145/3395035.3425184

M3 - Conference contribution

SN - 978-1-4503-8002-7

SP - 522

BT - ICMI '20 Companion

PB - Association for Computing Machinery

CY - New York

T2 - ACM International Conference on Multimodal Interaction

Y2 - 25 October 2020 through 29 October 2020

ER -

Automatic Recognition of Target Words in Infant-Directed Speech

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this