Development and Evaluation of Self-Regulation Electrolarynx Based on Mandarin Tone Characteristics

The electrolarynx (EL) is a primary device for voice restoration in patients who have undergone total laryngectomy. However, conventional monotonic EL (M-EL) typically produce a constant fundamental frequency (F0), resulting in flat, unnatural speech. This limitation is particularly detrimental for Mandarin Chinese speakers, where tones carry essential lexical meaning. Existing approaches to enhance F0 modulation—either via physiological signal control or speech signal processing—often involve complex hardware setups or suffer from suboptimal real-time performance, limiting their applicability. In this study, we proposed a novel self-regulating EL device (S-EL) capable of producing F0-modulated speech during phonation. Based on an analysis of the tone characteristics in Mandarin, an appropriate sinusoidal F0 contour was designed and customized of the target speaker. A prototype of S-EL was developed to implement this modulation scheme. Subjective evaluations demonstrated significant improvements in speech intelligibility and acceptability. Dictation accuracy for S-EL speech reached 67.32% ±27.53%, substantially higher than that of M-EL speech (48.27% ± 24.31%, p < 0.01). Mean opinion score (MOS) ratings for acceptability were also significantly better with the S-EL (3.04 ±0.66) compared to the M-EL (2.78 ±0.46, p < 0.05). Taken together, this work presents a promising technical pathway for improving the quality of tone language EL speech, and provides a more effective solution for voice rehabilitation in laryngectomees.