Towards Reducing Patient Effort for the Automatic Prediction of Speech Intelligibility in Head and Neck Cancers - Archive ouverte HAL Access content directly
Conference Papers Year : 2023

Towards Reducing Patient Effort for the Automatic Prediction of Speech Intelligibility in Head and Neck Cancers

Abstract

The automatic prediction of speech intelligibility can be seen as a growing and relevant alternative to the perceptual evaluations used clinically, which are known to be biased, variant and subjective. We propose an automatic way to regress an intelligibility score based on a recurrent model with a self-attention mechanism. This approach not only presented a high correlation of 0.87 when applied to a pseudo-word task designed for head and neck cancers, but also a significant decrease in error of more than 50%, when compared to previous approaches. Moreover, we have also studied the reliability of the same system when operating with smaller amounts of data at inference time. The results suggest that we can reduce the linguistic sample size to only 30% of the full sample, without losing performance. This aspect validates the reliability of using a smaller subset of data when predicting intelligibility, which can be extremely useful to prevent patient's fatigue by creating smaller batteries of clinical exams.
Fichier principal
Vignette du fichier
Towards_Reducing_Patient_Effort_for_the_Automatic_Prediction_of_Speech_Intelligibility_in_Head_and_Neck_Cancers.pdf (1.05 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04093771 , version 1 (10-05-2023)

Identifiers

Cite

Sebastião Quintas, Alberto Abad, Julie Mauclair, Virginie Woisard, Julien Pinquier. Towards Reducing Patient Effort for the Automatic Prediction of Speech Intelligibility in Head and Neck Cancers. 48th International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2023), IEEE Signal Processing Society (SPS), Jun 2023, Rhodes, Greece. pp.1-5, ⟨10.1109/icassp49357.2023.10094921⟩. ⟨hal-04093771⟩
16 View
8 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More