Acoustic analysis of tracheo-oesophageal voice: cepstral and voicing measures

Objectives: To develop an acoustic analysis protocol examining the correlations between acoustic and perceptual parameters in tracheo-oesophageal (TE) voices. Methods: Forty-nine laryngectomised patients with TE voice (42 M, 7 F; mean age 68 ± 9.3 years) were enrolled. Voice samples holding the sustained vowel /a/ and the Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V) sentences were recorded and analysed to extract the fundamental frequency (F0), the fraction of locally unvoiced frames (FLUF) and the cepstral peak prominence smoothed (CPPS). A perceptual evaluation was performed with the Impression, Noise, Fluency and quality of Voice (INFVo) scale. Results: CPPS and FLUF did not significantly correlate with F0. Average F0 on connected speech showed moderate correlations with I and Vo parameters (INFVo). Strong correlations between CPPS and FLUF were observed on both sustained vowel and connected speech. FLUF and CPPS moderately correlated with the I and Vo parameters; FLUF showed also a moderate correlation with the N parameter. Conclusions: CPPS and FLUF are evaluable on both sustained vowel and connected speech and are easy to analyse. The intercorrelation between them and the correlations with perceptual evaluation suggest their validity for TE voice assessment.