AI Paper Atlas

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

The Right To Talk: An Audio-Visual Transformer Approach

Thanh-Dat Truong, Chi Nhan Duong, The De Vu, Hoang Anh Pham, Bhiksha Raj, Ngan Le, Khoa Luu

ICCV 2021 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual source localization · 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 cross-domain alignment loss L_align (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 1S+NSDR (dB): 14.4 -> 15.8Improved
add cross-domain alignment loss L_align (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 2SSDR (dB): 9.8 -> 10.3Improved
add cross-domain alignment loss L_align (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 2S+NSDR (dB): 7.0 -> 7.2Improved
add cross-domain alignment loss L_align (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 2S+NPESQ: 2.5 -> 2.5No clear effect
add across-segment temporal attention (ablation)Audio-Visual Transformer (block attention) · Columbia (Avg) cooperative localizationF1 (%): 92.7% -> 93.4%Improved
add across-segment temporal attention (ablation)Audio-Visual Transformer (speaker attention) · Columbia (Avg) cooperative localizationF1 (%): 93.8% -> 94.9%Improved
add across-segment temporal attention (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 2SSDR (dB): 10.3 -> 11.5Improved
add across-segment temporal attention (ablation)Audio-Visual Transformer (block attention) · LRS2 separation 2S+NSDR (dB): 7.2 -> 8.1Improved
add across-segment temporal attention (ablation)Audio-only variant · LRS2 separation 2SSDR (dB): 9.1 -> 9.4Improved
add visual stream to the audio-only model (ablation)Audio-Visual Transformer · LRS2 separation 2SSDR (dB): 9.1 -> 10.3Improved
use speaker (face-detection) regions instead of block decomposition (ablation)Audio-Visual Transformer · LRS2 separation 1S+NSDR (dB): 15.8 -> 16.5Improved
use speaker regions instead of block decomposition (ablation)Audio-Visual Transformer · Columbia (Avg) cooperative localizationF1 (%): 92.7% -> 93.8%Improved

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