New study compares sEMG encoding accuracy across speech modes

A new research paper explores the effectiveness of Speech Articulatory Coding (SPARC) features for predicting surface electromyography (sEMG) envelopes across different speech modes. The study found that SPARC features provided higher prediction accuracy than traditional phoneme representations in aloud, mimed, and subvocal speech. These findings suggest SPARC is a robust intermediate target for developing silent-speech modeling technologies. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a more effective feature representation for sEMG-based silent-speech modeling.

RANK_REASON Academic paper on a novel feature representation for speech modeling.

Read on arXiv cs.CL →

• SPARC
• sEMG
• Chenqian Le

• paper
• other