Objective
This study addresses the need to teach the Vocal Profile Analysis Scheme protocol in the Brazilian Portuguese version (VPAS-PB) for use in Speaker Comparison (SC), highlighting the relevance of this tool as a potential resource for this purpose. The objective was to verify the applicability of VPAS-PB in Speaker Comparison and to evaluate the effectiveness of perceptual training based on this protocol with speech-language pathology students.

Methods
The research corpus consists of semi-spontaneous speech samples from ten male subjects who told significant stories from their lives, recorded simultaneously by tape recorder and cell phone. Initially, three speech therapists specialized in voice research and treatment with high inter-rater reliability in VPAS-PB described the vocal profiles of these samples. In the second stage, eighteen students participated in an eight-week course on the analysis of speakers' vocal profiles. The course covered adjustments of the lips, tongue, jaw, pharynx, velopharynx, and larynx height, as well as muscle tension, phonatory elements, and vocal dynamics. After the training, they performed a voice identification task on the vocal profiles established by the specialists. In the practical stage, students performed an exercise with 15 lineups using the participants' voices, which had been previously described by the experts. In the third stage, the perceptual strategies employed to describe the vocal profiles were analyzed based on the confusion matrix obtained from the responses.

Results
In the perceptual task performed after training, students predominantly used aspects of vocal dynamics as their main strategy for identifying voices. Analysis of the confusion matrix showed that the more salient the vocal dynamics traits, the higher the accuracy rate in matching the samples to the profiles described by the speech-language pathologists. Prosody was highlighted as a relevant perceptual element, often adopted as an initial recognition strategy. Overall, perceptual training proved to be feasible and effective, allowing most of the 18 participants to correctly identify voices based on the VPAS-PB criteria and demonstrating the instrument's potential for training in speaker comparison.

Conclusion
The VPAS-PB proved to be applicable as a complementary instrument in speaker comparison, and perceptual training based on its phonetic descriptors proved effective. These findings reinforce the potential of structured perceptual training programs to improve the accuracy of vocal analysis in clinical and forensic contexts.