Utterance-Level Methods for Identifying Reliable ASR-Output for Child Speech

Gus Lathouwers; Lingyun Gao; Catia Cucchiarini; Helmer Strik

eess.ASApr 10, 2026

Utterance-Level Methods for Identifying Reliable ASR-Output for Child Speech

Authors: Gus Lathouwers, Lingyun Gao, Catia Cucchiarini, Helmer Strik

Organizations: Centre for Language Studies, Radboud University, Netherlands · Department of Language and Communication, Radboud University, Netherlands

Abstract

Automatic Speech Recognition (ASR) is increasingly used in applications involving child speech, such as language learning and literacy acquisition. However, the effectiveness of such applications is limited by high ASR error rates. The negative effects can be mitigated by identifying in advance which ASR-outputs are reliable. This work aims to develop two novel approaches for selecting reliable ASR-output at the utterance level, one for selecting reliable read speech and one for dialogue speech material. Evaluations were done on an English and a Dutch dataset, each with a baseline and finetuned model. The results show that utterance-level selection methods for identifying reliably transcribed speech recordings have high precision for the best strategy (P > 97.4) for both read speech and dialogue material, for both languages. Using the current optimal strategy allows 21.0% to 55.9% of dialogue/read speech datasets to be automatically selected with low (UER of < 2.6) error rates.

Figures & tables

Figure 1: Audio Processing Pipeline for Different Strategies used.

Read material
	Dutch				English
Condition	P	R	F1	MCC	P	R	F1	MCC
Whisper-V2 [prompt]	80.7	60.8	69.3	0.51	97.8	74.0	84.3	0.28
Whisper-FT [prompt]	97.2	91.5	94.3	0.90	98.2	69.3	81.3	0.27
Agreement Whisper-V2 and Whisper-FT [prompt]	98.3	57.9	72.9	0.64	98.4	79.5	88.0	0.71
Dialogue material

Table 1: Performance Analysis of Methods for Selecting Reliable ASR-Output

Read material
	Dutch			English
Condition	of Dataset	UER	WER	of Dataset	UER	WER †
Whisper-V2	100.0%	56.2	16.7	100.0%	29.4	-
Whisper-V2 [prompt]	34.0%	19.3	5.2	70.6%	2.2	-
Whisper-FT	100.0%	24.3	5.1	100.0%	34.1	-
Whisper-FT [prompt]	42.1%	2.8	0.5	65.9%	1.8	-

Table 2: Subportion of Datasets Selected with Accompanying UER and WER

Explore similar work

Transcribing Children's Speech: ASR Performance and Obtaining Reliable Orthographic Transcriptions
Apr 10, 2026Gus Lathouwers, Lingyun Gao, Catia Cucchiarini +1Automatic Speech Recognition ASR Evaluation
Radboud University, the Netherlands
Enabling automatic transcription of child-centered audio recordings from real-world environments
Jun 13, 2025Daniil Kocharov, Azarias Galama, Okko RäsänenAutomatic Speech Recognition
Signal Processing Research Centre, Tampere University, Finland, Faculty of Information Technology and Communication Sciences, Department of Computing Sciences, P.O. Box 553, FI-33101 Tampere, Finland
Building an ASR Solution for Training and Assessing Children's Reading
Jun 30, 2026Yacouba Diarra, Nouhoum Souleymane Coulibaly, Mamadou Dembele +2Automatic Speech Recognition Low-Resource Speech Recognition
RobotsMali AI4D Laboratory Bamako, Mali

Apr 10, 2026J/K move · Enter open · S save

Transcribing Children's Speech: ASR Performance and Obtaining Reliable Orthographic Transcriptions

Gus Lathouwers, Lingyun Gao, Catia Cucchiarini, Helmer Strik

Radboud University, the Netherlands

Automatic Speech Recognition ASR Evaluation Low-Resource Speech Recognition

Open paper AskPDF arXiv

Automatic speech recognition (ASR) has the potential to substantially reduce manual annotation effort in child speech research by generating automatic transcriptions. However, obtaining reliably high-quality ASR transcriptions for child speech remains challenging in low-resource languages due to limited child-specific pre-trained models and highly diverse noise conditions. This study investigates the effectiveness of state-of-the-art ASR models on child speech through two research questions, by evaluating nine ASR models from three model families (Whisper, Parakeet, and Wav2Vec2) on two Dutch child speech datasets, JASMIN and DART. Research question 1 examines the performance of ASR-models applied to child speech. The fine-tuned Whisper-medium model achieves the best overall performance, with a WER of 5.54% on JASMIN and 70.37% on DART, showing that the noisy DART data are clearly more challenging. Research question 2 examines to what extent it is possible to select a subset for which reliable orthographic transcriptions can be obtained automatically, without the need for manual verification. We use an utterance-level selection method that compares ASR output with the original read prompt to identify correctly pronounced recordings. Using the proposed selection method, 42.0% [for JASMIN] and 18.1% [for DART] of the utterances can be automatically identified as correctly pronounced with high confidence, resulting in very low error rates on an utterance level (precisions of 98.3% and higher) and reducing the need for manual verification.

Abstract

Explore similar work

Transcribing Children's Speech: ASR Performance and Obtaining Reliable Orthographic Transcriptions

Enabling automatic transcription of child-centered audio recordings from real-world environments

Building an ASR Solution for Training and Assessing Children's Reading