Acoustic influence of the neck tissue on Mandarin voiceless consonant production of electrolaryngeal speech

Abstract

Lack of an appropriate voice source is the main reason for low intelligibility of the electrolaryngeal (EL) voiceless consonants. It is essential that the influence of neck tissue on EL voiceless consonant production is studied in order to design a suitable voice source. Firstly, the neck frequency response function was measured across a wide frequency range (100–20,000Hz) to investigate the potential impact of the neck tissue on the EL voiceless consonants. The results show that the low-pass characteristic of the neck tissue distorts the spectral shape of the EL voiceless consonant, due to increased energy attenuation at higher frequencies (>2259Hz). Then, a random and high-frequency energy-enhanced noise (HFEE-Noise) source was designed for EL voiceless consonant production. The results indicate that the HFEE-Noise source can compensate for the influence of the neck tissue, and shows promising potential for production of an EL voiceless consonant with a spectral shape that is closer to a natural voice. Furthermore, the results of perceptual experiments show that the HFEE-Noise source can effectively reduce the perceptual sonorization of the EL voiceless consonants, but shows limitations in increasing the perceptual discriminability of different voiceless consonants. Finally, some suggestions were proposed for the future design and implementation of an appropriate voice source to improve the EL voiceless consonant intelligibility.