VALLR-Pin: Uncertainty-Factorized Visual Speech Recognition for Mandarin with Pinyin Guidance

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Chang, Xie, Dongliang, Xie, Wanpeng, Qin, Bo, Yang, Hong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914223036039168
author Sun, Chang
Xie, Dongliang
Xie, Wanpeng
Qin, Bo
Yang, Hong
author_facet Sun, Chang
Xie, Dongliang
Xie, Wanpeng
Qin, Bo
Yang, Hong
contents Visual speech recognition (VSR) aims to transcribe spoken content from silent lip-motion videos and is particularly challenging in Mandarin due to severe viseme ambiguity and pervasive homophones. We propose VALLR-Pin, a two-stage Mandarin VSR framework that extends the VALLR architecture by explicitly incorporating Pinyin as an intermediate representation. In the first stage, a shared visual encoder feeds dual decoders that jointly predict Mandarin characters and their corresponding Pinyin sequences, encouraging more robust visual-linguistic representations. In the second stage, an LLM-based refinement module takes the predicted Pinyin sequence together with an N-best list of character hypotheses to resolve homophone-induced ambiguities. To further adapt the LLM to visual recognition errors, we fine-tune it on synthetic instruction data constructed from model-generated Pinyin-text pairs, enabling error-aware correction. Experiments on public Mandarin VSR benchmarks demonstrate that VALLR-Pin consistently improves transcription accuracy under multi-speaker conditions, highlighting the effectiveness of combining phonetic guidance with lightweight LLM refinement.
format Preprint
id arxiv_https___arxiv_org_abs_2512_20032
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VALLR-Pin: Uncertainty-Factorized Visual Speech Recognition for Mandarin with Pinyin Guidance
Sun, Chang
Xie, Dongliang
Xie, Wanpeng
Qin, Bo
Yang, Hong
Computer Vision and Pattern Recognition
Visual speech recognition (VSR) aims to transcribe spoken content from silent lip-motion videos and is particularly challenging in Mandarin due to severe viseme ambiguity and pervasive homophones. We propose VALLR-Pin, a two-stage Mandarin VSR framework that extends the VALLR architecture by explicitly incorporating Pinyin as an intermediate representation. In the first stage, a shared visual encoder feeds dual decoders that jointly predict Mandarin characters and their corresponding Pinyin sequences, encouraging more robust visual-linguistic representations. In the second stage, an LLM-based refinement module takes the predicted Pinyin sequence together with an N-best list of character hypotheses to resolve homophone-induced ambiguities. To further adapt the LLM to visual recognition errors, we fine-tune it on synthetic instruction data constructed from model-generated Pinyin-text pairs, enabling error-aware correction. Experiments on public Mandarin VSR benchmarks demonstrate that VALLR-Pin consistently improves transcription accuracy under multi-speaker conditions, highlighting the effectiveness of combining phonetic guidance with lightweight LLM refinement.
title VALLR-Pin: Uncertainty-Factorized Visual Speech Recognition for Mandarin with Pinyin Guidance
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.20032