Objective estimation of tracheoesophageal speech ratings using an auditory model

Area of Science:

• Otolaryngology
• Speech-Language Pathology
• Acoustic Analysis

Background:

• Total laryngectomy for advanced laryngeal cancer necessitates alternative voice production methods.
• Tracheoesophageal (TE) speech is a common post-laryngectomy voicing technique.
• TE speech often exhibits abnormal voice quality requiring objective assessment.

Purpose of the Study:

• To investigate the application of an auditory model for predicting subjective ratings of TE speech quality.
• To evaluate the model's ability to extract perceptually relevant acoustic features.
• To compare the model's performance against traditional telecommunication methods.

Main Methods:

• Utilized the Moore-Glasberg auditory model to extract acoustic features from TE speech.
• Combined extracted features to estimate subjective speech quality ratings.
• Validated the approach using speech samples from 35 TE speakers with existing subjective ratings.

Main Results:

• Achieved moderate correlations between objective acoustic metrics and subjective listener ratings.
• Demonstrated significantly improved correlations compared to traditional telecommunication quality estimation methods.
• Highlighted the potential of auditory models in quantifying TE speech quality.

Conclusions:

• Auditory models offer a promising approach for objective assessment of tracheoesophageal speech quality.
• This method can aid speech-language pathologists in TE speech rehabilitation.
• The findings suggest advancements in speech quality estimation for post-laryngectomy patients.