Abstract

FLOREZ-GOMEZ, Andrés Felipe; OROZCO-ARROYAVE, Juan Rafael  and  ROLDAN-VASCO, Sebastián. Correlation Between Speech-Related Feature Spaces and Clinical Voice Disorders in Patients with Dysphagia. TecnoL. [online]. 2022, vol.25, n.53, e204.  Epub Aug 08, 2022. ISSN 0123-7799.  https://doi.org/10.22430/22565337.2220.

Dysphagia is defined as the difficulty to transport an alimentary bolus from the oral cavity to the stomach in a safe and effective way. Currently, dysphagia-related diagnosis methods are invasive and highly dependent on the examiner’s experience. Biosignal-based studies, such as those on voice and speech records, have been proposed to develop complementary diagnostic tools. Likewise, this study explores, in features extracted from voice and speech signals, the capacity to discriminate between healthy subjects and patients with swallowing disorders. For this purpose, the signals were recorded in a group of 30 healthy individuals and 45 dysphagic patients. The participants performed different voice tasks (sustained vowels) and speech tasks (text reading, monologue, and diadochokinetic exercises). The patient records were assigned labels of three clinical conditions: wet voice, dysphonic voice, and voice with undetermined alteration. Classical voice- and speech-related feature spaces were assessed using statistical tests, and it was found that features related to phonation, prosody, and diadochokinesia have potential as biomarkers for the discrimination of different alterations in patients with dysphagia. This is a preliminary study based on voice and speech signals for a non-invasive and objective diagnosis of dysphagia.

Keywords : Dysphagia; Speech analysis; Voice analysis; Biosignal processing; Feature extraction; Statistical analysis.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )