JSPG: Dynamic Dictionary Filtering via Joint Semantic-Pinyin-Glyph Retrieval for Chinese Contextual ASR
Quick Answer
The JSPG framework enhances Chinese contextual ASR by integrating Semantic, Pinyin, and Glyph features, significantly improving keyword recognition accuracy on Aishell-1 and RWCS-NER datasets.
Quick Take
The JSPG framework enhances Chinese contextual ASR by integrating Semantic, Pinyin, and Glyph features, significantly improving keyword recognition accuracy on Aishell-1 and RWCS-NER datasets. This approach addresses homophonic errors that hinder traditional models, demonstrating superior performance over single-feature baselines.
Key Points
- JSPG integrates Semantic, Pinyin, and Glyph features for improved ASR accuracy.
- Significant performance gains observed on Aishell-1 and RWCS-NER datasets.
- Traditional models struggle with homophonic errors affecting semantic meaning.
- JSPG's extended Smith-Waterman algorithm bridges character and sequence-level filtering.
- Downstream ASR models guided by JSPG show substantial keyword recognition improvements.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Contextual Automatic Speech Recognition (ASR) faces challenges with large-scale keyword dictionaries, as excessive irrelevant candidates introduce noise that degrades accuracy. To address this, dynamic filtering typically uses a base ASR model to generate preliminary hypotheses, followed by semantic text retrievers to fetch a concise subset of relevant keywords. However, this approach frequently fails in Chinese ASR. Base models often produce homophonic or near-homophonic errors that preserve the phonetic cues of the target keywords but severely distort their semantic meaning, rendering standard semantic retrievers ineffective. To resolve this, we propose a filtering framework that jointly integrates Semantic, Pinyin, and Glyph features (JSPG). Pinyin effectively retrieves targets based on phonetic similarity, while glyph provides complementary structural cues to filter out numerous irrelevant homophones inherent in Chinese. To bridge the gap between character-level pinyin/glyph metrics and sequence-level filtering, we introduce an extended Smith-Waterman algorithm that computes similarity scores between the N-best hypothesis sequences and keywords. Experiments on the Aishell-1 and RWCS-NER datasets demonstrate that JSPG significantly outperforms single-feature baselines. Furthermore, downstream contextual ASR models guided by JSPG achieve substantial improvements in keyword recognition accuracy.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2605.16896 [cs.CL] |
| (or arXiv:2605.16896v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.16896 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Shilin Zhou [view email]
[v1]
Sat, 16 May 2026 09:16:09 UTC (72 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 →Quantifying Prior Dominance in Systems
The study introduces the Normalized Context Utilization (NCU) metric to evaluate Retrieval-Augmented Generation (RAG) systems, revealing that Small Language Models (SLMs) outperform larger models in factual extraction. The findings indicate that traditional scaling laws yield diminishing returns, with a commercial API frequently failing against adversarial evidence due to systemic confidence collapse.