JSPG: Dynamic Dictionary Filtering via Joint Semantic-Pinyin-Glyph Retrieval for Chinese Contextual ASR

2026-05-16 · Shilin Zhou, Zhenghua Li arxiv

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.

📄 PDF Abstract BibTeX arXiv:2605.16896

Code (0)

등록된 구현이 없습니다.

Tasks

Speech Recognition

Spatial Transformer Point Convolution

2020-09-03 · Yuan Fang, Chunyan Xu, Zhen Cui, Yuan Zong 외

Point clouds are unstructured and unordered in the embedded 3D space. In order to produce consistent responses under different permutation layouts, most existing methods aggregate local spatial points through maximum or …

Dictionary Learning Semantic Segmentation

Accessing and Elaborating Walenty - a Valence Dictionary of Polish - via Internet Browser

2016-05-01 · LREC 2016 5 · Bart{\l}omiej Nito{\'n}, Tomasz Bartosiak, El{\.z}bieta Hajnicz

This article presents Walenty - a new valence dictionary of Polish predicates, concentrating on its creation process and access via Internet browser. The dictionary contains two layers, syntactic and semantic. The syntac…

TsallisPGD: Adaptive Gradient Weighting for Adversarial Attacks on Semantic Segmentation

2026-05-05 · Alexander Matyasko, Xin Lou, Indriyati Atmosukarto, Wei Zhang arxiv

Attacking semantic segmentation models is significantly harder than image classification models because an attacker must flip thousands of pixel predictions simultaneously. Standard pixel-wise cross-entropy (CE) is ill-s…

Semantic Segmentation Image Classification Adversarial Attack

Joint Dictionaries for Zero-Shot Learning

2017-09-12 · Soheil Kolouri, Mohammad Rostami, Yuri Owechko, Kyungnam Kim

A classic approach toward zero-shot learning (ZSL) is to map the input domain to a set of semantically meaningful attributes that could be used later on to classify unseen classes of data (e.g. visual data). In this pape…

Attribute Dictionary Learning Zero-Shot Learning

Low-Rank Embedded Ensemble Semantic Dictionary for Zero-Shot Learning

2017-07-01 · CVPR 2017 7 · Zhengming Ding, Ming Shao, Yun Fu

Zero-shot learning for visual recognition has received much interest in the most recent years. However, the semantic gap across visual features and their underlying semantics is still the biggest obstacle in zero-shot le…

Dictionary Learning Zero-Shot Learning