Research map / Video / Video-language understanding
Lip Reading Sentences in the Wild
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Add curriculum learning, starting with single-word examples and gradually increasing sequence length (ablation) | WAS · LRS test, lips only | WER: 76.5% -> 61.1% | Helped |
| Add scheduled sampling to WAS+CL (ablation) | WAS · LRS test, lips only | WER: 61.1% -> 58.1% | Helped |
| Add beam search to WAS+CL+SS (ablation) | WAS · LRS test, lips only | WER: 58.1% -> 50.2% | Helped |
| Add lip input to audio-only recognition | LAS+CL+SS+BS · LRS test, clean audio | CER: 10.4% -> 7.9% | Helped |
| Add lip input to audio-only recognition | LAS+CL+SS+BS · LRS test, 10dB audio | WER: 37.6% -> 27.6% | Helped |
| Add lip input to audio-only recognition | LAS+CL+SS+BS · LRS test, 0dB audio | WER: 62.9% -> 42.0% | Helped |
| Use a bidirectional rather than unidirectional encoder (ablation) | WLAS · LRS | performance | No clear effect |
| Remove attention (ablation) | WLAS · LRS | WER: over 100% | Hurt |
| Increase beam width beyond 4 (ablation) | WLAS decoding · LRS test | WER | No clear effect |
| Use WAS instead of the professional lip reader | LRS test, lips only · LRS test | WER: 73.8% -> 50.2% | Helped |
Closest papers by meaning
- Personalized Lip Reading: Adapting to Your Unique Lip Movements with Vision and Language (AAAI 2025 · similarity 0.92)
- LipVoicer: Generating Speech from Silent Videos Guided by Lip Reading (ICLR 2024 · similarity 0.91)
- Language Without Borders: A Dataset and Benchmark for Code-Switching Lip Reading (NeurIPS 2024 · similarity 0.90)
- Let There Be Sound: Reconstructing High Quality Speech from Silent Videos (AAAI 2024 · similarity 0.86)
- Learning Visual Speech Representations via Cross-Modal Distillation and Joint Face-Lip Modeling (NeurIPS 2026 · similarity 0.86)
- MTGA: Multi-View Temporal Granularity Aligned Aggregation for Event-Based Lip-Reading (AAAI 2025 · similarity 0.85)
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.