Which Languages Transfer Best to Warlpiri? A Similarity-Based Study for Low-Resource ASR
Quick Answer
This study reveals that acoustic and typological similarities significantly enhance ASR performance for Warlpiri, a low-resource language.
Quick Take
This study reveals that acoustic and typological similarities significantly enhance ASR performance for Warlpiri, a low-resource language. Using Whisper, Assamese and Hindi showed notable reductions in word and character error rates, with acoustic similarity being the strongest predictor of fine-tuning success. The findings underscore the importance of language transfer in low-resource settings.
Key Points
- Acoustic and typological similarities improve ASR for Warlpiri significantly.
- Whisper model experiments showed Assamese and Hindi reduced error rates substantially.
- Acoustic similarity is the strongest predictor for fine-tuning performance.
- Phoneme inventory and typological similarity aid in zero-shot transfer.
- Research accepted by Interspeech 2026 highlights low-resource ASR challenges.
Paper Resources
📖 Reader Mode
~2 min readAbstract:This paper investigates how language similarity can improve cross-lingual transfer for automatic speech recognition (ASR) in extremely low-resource settings. Warlpiri, an Australian Aboriginal language, has very limited transcribed speech data, making transfer learning essential. We propose a framework combining acoustic similarity from pre-trained speech models with linguistic similarity based on typology, phoneme inventories, grammatical, and syntactic features to rank high-resource source languages and evaluate their effectiveness for ASR transfer to Warlpiri. Experiments with Whisper show that acoustically and typologically similar languages outperform monolingual and multilingual baselines. Assamese and Hindi achieve substantial reductions in word and character error rates. Correlation analysis further indicates that acoustic similarity is the strongest predictor of fine-tuning performance, while phoneme inventory and typological similarity better explain zero-shot transfer.
| Comments: | Accepted by Interspeech 2026 |
| Subjects: | Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) |
| Cite as: | arXiv:2607.10256 [cs.CL] |
| (or arXiv:2607.10256v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.10256 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Pravina Mylvaganam [view email]
[v1]
Sat, 11 Jul 2026 10:57:24 UTC (1,112 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 →Letting the Data Speak: Extracting Keywords from Crowdsourced Collections with AI
The study evaluates three NLP approaches—Named Entity Recognition, Keyword Extraction, and Topic Modelling—using the Their Finest Hour Online Archive to automate keyword extraction from crowdsourced WWII collections. Findings suggest that while NLP methods show promise, no single approach is sufficient, and ethical considerations in automated keyword extraction are crucial for responsible stewardship.