aiqbench analysis
This empirical study addresses the critical challenge of adapting Automatic Speech Recognition models for children without degrading performance on adult speech, a key issue in inclusive AI deployment. The research evaluates various fine-tuning strategies across Arabic and English datasets, demonstrating that weight-space merging techniques like LERP and TIES effectively balance child adaptation gains with adult retention. For practitioners building accessible voice interfaces, the findings highlight bilingual adaptation as a more stable approach than language-specific tuning when managing this specific trade-off.
Child ASR Adaptation with Adult Retention: An Empirical Study
arXiv:2610.08827v1 Announce Type: new Abstract: Automatic Speech Recognition (ASR) systems often underperform for children and non-native speakers, while adapting adult ASR models to child speech can cause adult-speech forgetting. We study child ASR adaptation with adult retention across Arabic and English. We compare full fine-tuning, LoRA, and post-hoc weight-space merging across encoder--decoder, encoder--CTC, and AudioLLM-based ASR systems. Experiments use Arabic native and non-native child speech, English MyST child speech, and adult benchmarks from MGB-2 and LibriSpeech test-clean. We evaluate recognition quality with WER and quantify the adaptation--retention trade-off using Retention Index, Child Adaptation Gain, and Adaptation Recovery. Results show that child adaptation is necessary, especially for non-native Arabic and English child speech, but direct adaptation often reduces adult ASR performance. Bilingual adaptation is more stable than language-specific adaptation. Weight-space merging often improves the trade-off, especially for encoder--CTC, Whisper, and AudioLLM-based ASR, with LERP favoring adult retention and TIES recovering stronger child gains. For the encoder--decoder model, direct bilingual fine-tuning remains strongest in raw WER.\footnote{Code, and models are available at https://github.com/qcri/Child-ASR-Adaptation.
Abstract:Automatic Speech Recognition (ASR) systems often underperform for children and non-native speakers, while adapting adult ASR models to child speech can cause adult-speech forgetting. We study child ASR adaptation with adult retention across Arabic and English. We compare full fine-tuning, LoRA, and post-hoc weight-space merging across encoder--decoder, encoder--CTC, and AudioLLM-based ASR systems. Experiments use Arabic native and non-native child speech, English MyST child speech, and adult benchmarks from MGB-2 and LibriSpeech test-clean. We evaluate recognition quality with WER and quantify the adaptation--retention trade-off using Retention Index, Child Adaptation Gain, and Adaptation Recovery. Results show that child adaptation is necessary, especially for non-native Arabic and English child speech, but direct adaptation often reduces adult ASR performance. Bilingual adaptation is more stable than language-specific adaptation. Weight-space merging often improves the trade-off, especially for encoder--CTC, Whisper, and AudioLLM-based ASR, with LERP favoring adult retention and TIES recovering stronger child gains. For the encoder--decoder model, direct bilingual fine-tuning remains strongest in raw WER.\footnote{Code, and models are available at this https URL.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS)
arXiv:2610.08827 [cs.CL]
(or arXiv:2610.08827v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.08827 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 273 words
Related coverage
AI Companion Robots Are Closing the Human Connection in Modern Homes
IEEE Spectrum · Aug 25, 2026
Digital States 2026: AI Moves From Experiment to Infrastructure
govtech.com · Sep 24, 2026
Spectral Rank Certification for Foundation Model Adapters
arXiv · Aug 19, 2026
A research agenda for the Economic Futures Research Fund
Anthropic · Jul 22, 2026