A semi-automatic pipeline for transcribing and segmenting child speech

Tools

Christodoulidou, Polychronia and Tanner, James and Stuart-Smith, Jane and McAuliffe, Michael and Murali, Mridhula and Smith, Amy and Taylor, Lauren and Cleland, Joanne and Kuschmann, Anja; (2025) A semi-automatic pipeline for transcribing and segmenting child speech. In: Proceedings of Interspeech 2025. Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH . ISCA Archive, NLD, pp. 4278-4282. (https://www.isca-archive.org/interspeech_2025/chri...)

[thumbnail of Christodoulidou-etal-Interspeech-2025-A-semi-automatic-pipeline-for-transcribing-and-segmenting-child-speech]
Preview
 Text. Filename: Christodoulidou-etal-Interspeech-2025-A-semi-automatic-pipeline-for-transcribing-and-segmenting-child-speech.pdf
Final Published Version
License: Other
Download (558kB)| Preview

Abstract

This study evaluates both automated transcription (WhisperX) and forced alignment (MFA) in developing a semi-automated pipeline for obtaining acoustic vowel measures from field recordings from 275 children speaking a non-standard, English dialect, Scottish English. As expected, manual correction of speech transcriptions before forced alignment improves the quality of acoustic vowel measures with respect to manually-annotated data, though speech style and recording environment present some challenges for both tools. Adaptation of the MFA pre-trained english_us_arpa acoustic model towards the children's speech also improves the quality of acoustic measures, though greater improvement was not found by increasing training sample size.

ORCID iDs

Christodoulidou, Polychronia, Tanner, James, Stuart-Smith, Jane, McAuliffe, Michael, Murali, Mridhula ORCID logoORCID: https://orcid.org/0000-0001-5450-6419, Smith, Amy ORCID logoORCID: https://orcid.org/0009-0001-6303-0691, Taylor, Lauren, Cleland, Joanne ORCID logoORCID: https://orcid.org/0000-0002-0660-1646 and Kuschmann, Anja ORCID logoORCID: https://orcid.org/0000-0001-5396-9008;
CORE (COnnecting REpositories)