AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Deep Audio-Visual Speech Recognition

Triantafyllos Afouras, Joon Son Chung, Andrew Senior, Oriol Vinyals, Andrew Zisserman

TPAMI 2018 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual speech separation · 863 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 audio to video input (AV vs V)TM-seq2seq · LRS2-BBC noisy audioWER: 58.0% -> 35.9%Improved
add video to audio input on clean audioTM-CTC · LRS2-BBC cleanWER: 10.1% -> 8.2%Improved
add video to audio input on clean audioTM-seq2seq · LRS2-BBC cleanWER: 9.4% -> 8.5%Improved
add external language modelTM-CTC · LRS2-BBC video-onlyWER: 65.0% -> 54.7%Improved
add external language modelTM-seq2seq · LRS2-BBC video-onlyWER: 49.8% -> 48.3%Improved
use seq2seq instead of CTC for lip readingTM-seq2seq · LRS2-BBC video-onlyWER: 54.7% -> 48.3%Improved
use CTC instead of seq2seq under loud noiseTM-CTC · LRS2-BBC noisy audio-visualWER: 34.2% -> 23.6%Improved
train with babble noise at probability 0.25audio-visual models · trainingWERImproved
introduce curriculum learning from single words to full sentencesTM models · trainingconvergenceImproved
fine-tune on randomly shifted videoTM-seq2seq · LRS2-BBC out-of-syncWERImproved
fine-tune on randomly shifted videoTM-CTC · LRS2-BBC out-of-syncWERNo clear effect
increase beam widthTM-seq2seq · LRS2-BBC video-onlyWERImproved

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