RESUMO
OBJECTIVE: This study aimed to present an experience in rating voices of adults with normal hearing and adults with cochlear implants and critically examine the outcomes, discussing pros and cons of the methodology used. STUDY DESIGN: This is a cross-sectional, prospective study. METHODS: One hundred and fifty voice samples, consisting of 50 sustained vowels, 50 samples of connected speech, and 50 samples of conversational speech, belonging to 25 adults with hearing impairment with cochlear implants and 25 adults with normal hearing, were perceptually analyzed for inter-rater agreement and intra-rater reliability. Three experienced judges rated the voice samples using visual analog scales of parameters considered relevant for cochlear-implanted population such as articulation, intonation, and resonance. The raters participated in three training sessions for calibration and had 1 month to complete the ratings individually. Twenty percent of the samples were repeated randomly to verify intra-rater reliability. The levels of agreement and reliability were verified using the interclass correlation coefficient. RESULTS: The inter-rater agreement varied widely across the parameters and speech tasks, from poor to excellent agreement. The only parameter for which the raters maintained consistently good or excellent agreement for all groups and emissions was the pitch. For intra-rater reliability, two of the raters presented excellent reliability for most parameters across all of the speech tasks, whereas one rater presented more inconsistencies. CONCLUSIONS: In this reliability study, factors such as extensive deadline for the auditory perceptual evaluation, lack of periodic recalibration, speech tasks, and familiarity with the population studied were identified as factors that contributed to inconsistent reliability results.
