Related Experiment Video
Updated: Jul 31, 2025
12:43
A Protocol for Comprehensive Assessment of Bulbar Dysfunction in Amyotrophic Lateral Sclerosis ALS
Published on: February 21, 2011
34.9K
MagTrack: A Wearable Tongue Motion Tracking System for Silent Speech Interfaces
Beiming Cao1,2, Shravan Ravi3, Nordine Sebkhi4
1Department of Electrical and Computer Engineering, The University of Texas at Austin.
Summary
MagTrack, a new wearable electromagnetic articulograph, shows promise for silent speech interfaces. It achieved high accuracy in vowel classification and comparable results to commercial devices in continuous speech recognition.
Area of Science:
- Biomedical Engineering
- Speech Technology
- Wearable Devices
Background:
- Current electromagnetic tongue tracking devices are limited for daily use, hindering applications like silent speech interfaces.
- A novel wearable electromagnetic articulograph, MagTrack, has been developed to address these limitations.
Purpose of the Study:
- To validate the performance of MagTrack for potential silent speech interface applications.
- To compare MagTrack's efficacy against commercial electromagnetic articulograph systems.
Main Methods:
- Two experiments were conducted: isolated vowel classification and continuous silent speech recognition using MagTrack data from healthy adult speakers.
- Performance metrics included classification accuracy for vowels and phoneme error rates for continuous speech.
- Results were benchmarked against data from a commercial electromagnetic articulograph.
Main Results:
- MagTrack achieved 89.74% accuracy in isolated vowel classification, outperforming previous results with commercial devices.
- Continuous silent speech recognition yielded phoneme error rates of 73.92% and 66.73% for two subjects.
- MagTrack demonstrated comparable performance to commercial systems when using similar localized information.
Conclusions:
- MagTrack shows potential for silent speech interfaces as a lightweight, wearable device.
- Incorporating raw magnetic signals could further enhance MagTrack's performance.
- This technology may also support speech therapy and language learning applications.
