Child ASR Adaptation with Adult Retention: An Empirical Study
Quick Answer
This study investigates child ASR adaptation while retaining adult performance, revealing that bilingual adaptation is more stable than language-specific approaches.
Quick Take
Techniques like weight-space merging enhance trade-offs, particularly for encoder-CTC and AudioLLM systems, while direct adaptation often hampers adult ASR performance. Results indicate that child adaptation is crucial, especially for non-native speakers in Arabic and English.
Key Points
- Child ASR adaptation is necessary for non-native Arabic and English speakers.
- Bilingual adaptation shows greater stability compared to language-specific methods.
- Weight-space merging improves adaptation-retention trade-offs in ASR systems.
- Direct adaptation often reduces performance of adult ASR models.
- Encoder-CTC and AudioLLM benefit significantly from advanced adaptation techniques.
Paper Resources
📖 Reader Mode
~2 min readAbstract: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.
| Comments: | long paper |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS) |
| MSC classes: | 68T50 |
| ACM classes: | F.2.2; I.2.7 |
| Cite as: | arXiv:2610.08827 [cs.CL] |
| (or arXiv:2610.08827v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.08827 arXiv-issued DOI via DataCite |
Submission history
From: Houssam Eddine-Othman Lachemat [view email]
[v1]
Fri, 25 Sep 2026 18:10:54 UTC (651 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.