Klaske van Sluis

82 5.2. Speech and methods from the lungs to the mouth is thus reestablished. Henceforth, the patient is able to produce pulmonary driven speech again. The new voice source is the pharyngoesophageal segment (PES). Tracheoesophageal speech is considered the gold standard in restoring com- municative functioning after TL [1]. It is considered as the most natural way of voice restoration according to intelligibility, pitch, and range [4]. Success rates for acquiring tracheoesophageal speech are reported up to 95% [5]. The reached endpoint in voice quality and speech intelligibility varies between pa- tients [1]. Effective vibratory functioning of the PES is crucial in acquiring tracheoesophageal speech. Knowledge about long-term voice outcomes of tra- cheoesophageal speakers so far is scarce. There are some studies that include evaluations of tracheoesophageal speakers on the long-term, up to 18 years post-TL [6–10]. These papers, however, do not evaluate the groups of patients by follow-up time [6, 7, 9, 10]. Studies which consider long-term follow-up thus far only assess communication mode and quality of life [4]. In voice and speech assessment, a multidimensional approach is preferred. Acoustic, perceptual, aerodynamic, stroboscopic and self-assessment can be used to evaluate voice quality [11]. Substitute voices characteristically deviate from healthy speakers because of strong voice irregularities and require a well- thought-out approach [12]. As communication is mostly a perceptual matter, perceptual evaluations are considered the “gold standard”. For substitute voices, judgments of experienced speech-language pathologists (SLP’s) are considered as more consistent than judgments of naïve raters [9]. Various perceptual scales are applied in the literature to rate substitute voices. The IINFVo rating scale was specifically developed for substitute speech [13]. The five IINFVo scale pa- rameters are: overall impression (I), impression of intelligibility (I), unintended additive noise (N), fluency (F) and voicing (Vo) [14]. Two of the rating scales, over-all impression and intelligibility, appear to be the most reliable [13–15] and are used in this study. The former refers to the acceptability or pleasantness of the voice (voice quality) and the latter refers to the clarity and understandabil- ity of words and sentences [14, 15]. During the last decade, automatic speech and voice analysis became feasible [16, 17]. Automatic analysis is promising in providing consistent ratings and for analyzing trends within a single speaker. The present study aims to identify changes in tracheoesophageal speech over time by analyzing perceptual and automatic evaluations of voice recordings. 5.2 Speech and methods 5.2.1 Speakers and recordings The Netherlands Cancer Institute has a long history of speech collection for TL research. Recordings from 13 TL-patients, who participated in studies be- tween 1996 and 2014, are included in the present study (all male, median age at treatment 55 years, range 44-75, all gave informed consent). Inclusion was