Text Entry via Discrete and Analog Myoelectric Signals
Adam J. Sporka, Antonín Posusta, Ondrej Poláček, Tomáš Flek, Jakub Otáhal · 2014 · ASSETS '14: Proceedings of the 16th International ACM SIGACCESS Conference on Computers & Accessibility · doi:10.1145/2661334.2661426
Summary
This demo paper presents a prototype system for text entry using surface electromyography (sEMG) signals detected from muscles in the user's forearm. The system is intended for people with upper extremity disabilities who retain the ability to control their forearm muscles. Using standard ECG-type self-adhesive electrodes placed above the extensor muscles on the forearm (extensor carpi ulnaris and extensor digitorum), a custom-made portable USB device captures EMG signals, which are processed through an analog amplifier and filter stage and digitized by an ARM microprocessor. The system can detect movements of two individual fingers (typically index and little finger) and the entire palm, producing both discrete binary signals (muscle active/inactive) and continuous analog quantifications of exerted force.
Key findings
The prototype implements three text entry methods across two input modes. In discrete mode: (1) 2-FOCL, adapted from the LetterWise method, displays letters in a 3x3 matrix where groups of 3-4 letters are selected by two finger movements, with letter probability prediction reordering letters to reduce keystrokes; and (2) Scanning LetterWise, where a cursor scans through letter groups and the user selects with a single muscle contraction. In continuous/analog mode: the system uses the quantified force of palm movement to control a pointer that selects from a ring of letter groups, with force magnitude mapping to pointer position. This analog approach exploits the continuous nature of EMG signals rather than reducing them to binary inputs, potentially offering faster text entry by allowing proportional control.
Relevance
This work explores an important direction in assistive input technology: using biosignals from residual muscle activity for computer interaction. For people who cannot use conventional keyboards or touchscreens but retain some forearm muscle control, EMG-based text entry offers a non-invasive alternative that requires minimal physical movement. The distinction between discrete and analog EMG processing is significant—while most assistive switch interfaces reduce input to binary on/off signals, this system's analog mode preserves the richness of the muscle signal, potentially enabling more expressive and efficient interaction. For assistive technology developers, the use of standard ECG electrodes and a portable USB device demonstrates that EMG-based input can be implemented with relatively low-cost, commercially available components.
Tags: assistive technology · text entry · electromyography · biosignals · motor impairments · input devices