AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Lip Reading for Low-resource Languages by Learning and Combining General Speech Knowledge and Language-specific Knowledge

Minsu Kim, Jeong Hun Yeo, Jeongsoo Choi, Yong Man Ro

ICCV 2023 · Large language models › Language, safety and interpretability · approach: Sign and cross-lingual translation · 37 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add LMDecoder (general + language-specific knowledge) on EnglishAV-HuBERT · LRS2WER (%): 25.5 -> 23.8Improved
pre-train visual encoder on speech-unit prediction instead of supervised English lip readingCM-seq2seq · mTEDx-FRWER (%): 88.41 -> 77.36Improved
pre-train decoder on target-language audio-text dataAV-HuBERT · mTEDx-ESWER (%): 71.68 -> 70.80Improved
use LMDecoder with memory instead of plain ASR-pretrained decoderASR Pre-train · mTEDx-ITWER (%): 71.28 -> 68.04Improved
add Language-specific Memory (ablation)LMDecoder · mTEDx-ITWER (%): 71.01 -> 68.04Improved
add MLS-IT audio-text data for LMDecoder (ablation)LMDecoder · mTEDx-ITWER (%): 73.24 -> 70.45Improved
combine MLS-IT and mTEDx-IT audio-text data (ablation)LMDecoder · mTEDx-ITWER (%): 70.45 -> 68.04Improved
reduce video-text data to 15.7h (ablation)Proposed Method · mTEDx-ITWER (%): 68.04 -> 75.62Hurt
use only 12h audio-text data for LMDecoder (ablation)Proposed Method · mTEDx-ITWER (%): 73.24 -> 86.7Hurt
increase audio-text data from 147h to 294h (ablation)Proposed Method · mTEDx-ITWER (%): 70.3 -> 68.0Improved
LMDecoder ASR performance vs plain ASR pre-trainLMDecoder · mTEDx-IT (ASR)WER (%): 24.65 -> 29.21Hurt
add LMDecoder to 3448h-English-pretrained encoderCM-seq2seq · mTEDx-ESWER (%): 59.90 -> 56.96Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy