ML Research Wiki / Benchmarks / Visual Speech Recognition / LRS3-TED

LRS3-TED

Visual Speech Recognition Benchmark

Performance Over Time

📊 Showing 3 results | 📏 Metric: Word Error Rate (WER)

Top Performing Models

Rank Model Paper Word Error Rate (WER) Date Code
1 VTP 📚 Sub-word Level Lip Reading With Visual Attention 40.60 2021-10-14 -
2 VTP with more data 📚 Sub-word Level Lip Reading With Visual Attention 30.70 2021-10-14 -
3 CTC/Attention 📚 Auto-AVSR: Audio-Visual Speech Recognition with Automatic Labels 19.10 2023-03-25 📦 mpc001/auto_avsr 📦 umbertocappellazzo/llama-avsr

All Papers (3)