Silent Speech Recognition on SOTA sensing-based SSI datasets (various)
93AccuracyEarSSR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EarSSRSensing Signal=Acoustic, Wear or hold devices=Earphone, DataSet=50 words and phrase, Recognition Scope=Closed-set (pre-defined word)2026.01 | 93 | — | |
| LipReaderSensing Signal=Acoustic, Wear or hold devices=Smartphone, DataSet=50 words, Recognition Scope=Closed-set (pre-defined word)2026.01 | 91.58 | — | |
| SoundLipSensing Signal=Acoustic, Wear or hold devices=Smartphone, DataSet=20 words and 70 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | 91.2 | — | |
| MuteItSensing Signal=IMU, Wear or hold devices=Earphone, DataSet=100 words, Recognition Scope=Closed-set (pre-defined word)2026.01 | 91 | — | |
| CELIPSensing Signal=Acoustic, Wear or hold devices=VR headsets, DataSet=70 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | 90.8 | — | |
| HearMeSensing Signal=Radio-frequency, Wear or hold devices=RFID Tag, DataSet=20 words, Recognition Scope=Closed-set (pre-defined word)2026.01 | 88 | — | |
| RFID TattooSensing Signal=Radio-frequency, Wear or hold devices=RFID Tag, DataSet=100 words and 20 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | 86 | — | |
| Lip-SiriSensing Signal=Wi-Fi, Wear or hold devices=None, DataSet=3398 words and 340 sentences, Recognition Scope=Open-sentence (Lexicon-based contextual decoding)2026.01 | 85.61 | — | |
| E-MASKSensing Signal=Strain sensor, Wear or hold devices=Mask with sensor, DataSet=21 commands, Recognition Scope=Closed-set (pre-defined word)2026.01 | 84.4 | — | |
| EchoLipSensing Signal=Acoustic, Wear or hold devices=Smartphone, DataSet=500 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | — | 13.9 | |
| LipwatchSensing Signal=Acoustic & IMU, Wear or hold devices=Smartwatch, DataSet=65 words and 80 commands, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | — | 13.7 | |
| mSilentSensing Signal=mmWave radar, Wear or hold devices=None, DataSet=1000 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | — | 9.5 | |
| TWLip-SeqSensing Signal=4-D Radar, Wear or hold devices=None, DataSet=30 sentences, Recognition Scope=Closed-set (pre-defined sentence)2026.01 | — | 4.55 |