Digital Voicing of Silent Speech
This work addresses the challenge of enabling communication for individuals who cannot vocalize, such as those with speech impairments, by improving silent speech synthesis, though it is incremental as it builds on prior EMG-based methods.
The paper tackles the problem of converting silently mouthed words into audible speech using EMG sensor measurements, achieving a reduction in transcription word error rate from 64% to 4% in one condition and 88% to 68% in another by training on silent EMG with transferred audio targets.
In this paper, we consider the task of digitally voicing silent speech, where silently mouthed words are converted to audible speech based on electromyography (EMG) sensor measurements that capture muscle impulses. While prior work has focused on training speech synthesis models from EMG collected during vocalized speech, we are the first to train from EMG collected during silently articulated speech. We introduce a method of training on silent EMG by transferring audio targets from vocalized to silent signals. Our method greatly improves intelligibility of audio generated from silent EMG compared to a baseline that only trains with vocalized data, decreasing transcription word error rate from 64% to 4% in one data condition and 88% to 68% in another. To spur further development on this task, we share our new dataset of silent and vocalized facial EMG measurements.