Voice Biometric System: The Identification of the Severity of Cerebral Palsy using Mel-Frequencies Stochastics Approach

Abstract

Cerebral Palsy (CP) is known as one type of disabilities that affects children. CP is an 'umbrella' for all chronic neurologic disorders which was manifested in movement control disorders. They arise early in life, commonly initiated with non-progressive diseases [1]. This neurological condition causes problems in body movement and muscle control that can inhibit the development of children's speech. The inhibition of speech development in CP is caused by the stiffness of the muscles around the mouth and face. It causes the children with CP having difficulty coordinating their speech organs. In CP, there is a disorder occurs that causes the muscles to move on their own without realizing it. This disorder is called dystonia. Regarding the development of speech on CP where the development is affected by the stiffness of the muscles around the mouth, then dystonia can be an indication of the severity level of CP. Based on dystonia that occurs on CP, CP can be classified into 4 severity levels: Slight, Mild, Moderate, and Severe. In Slight level, dystonia occurs less than 10% of the time and doesn't interfere with speech. Whereas in mild level of severity, dystonia occurs less than 50% and doesn't interfere with the speech. In moderate level, speech impairments caused by dystonia in CP begin to suffer where the dystonia is more than 50% of the time. Even at severe level, children with CP are difficult to move his/her mouth [2],[3]. Dystonia that occurs in CP causes CP speech has unique characteristics like "slurred," "choppy," or "mumbled" that may be difficult to understand. CP speech is also at a slow rate and the rapid rate in mumbling quality. Voice quality changes in CP, such as hoarse and breathy voice or speech that sounds "nasal" or "stuffy", are caused by limited tongue,

Voice as a Biomarker of Pediatric Health: A Scoping Review

“…India [7] Aggarwal 2020a [13] Aggarwal 2018 [14] Aggarwal 2020 [15] Dubey 2018 [53] Jayasree 2021 [9] Moharir 2017 [62] Sharma 2022 [31] Indonesia [4] Amrulloh 2015 [39] Amrulloh 2015 [43] Amrulloh 2018 [40] Nafisah 2019 [69] Italy [1] Tartarisco 2021 [56] Japan [1] Nakai 2017 [24] Lebanon [1] Salehian Matikolaie 2020 [68] Malaysia [1] Hariharan 2018 [10] Palestine [1] Khalilzad 2022 [66] Poland [4] Kotarba 2020…”

Forest quality assessment based on bird sound recognition using convolutional neural networks

“…In this paper, the system firstly processes the audio data into a spectrogram. The model with time-domain spectrogram data is then converted to the form of mel-frequency cepstral coefficients (MFCC) [27] or mel-spectrogram via the Fourier transform [28], as shown in Figure 2. One of the CNN models with good performance is the Wide-ResNet model [7].…”

Hand gesture-based automatic door security system using squeeze and excitation residual networks

“…In connection with efforts to prevent the spread of contagious diseases such as COVID-19 during the pandemic and to avoid the spread of infection due to the virus, using security systems in direct contact should be avoided [1]. Some security systems use biometric methods such as voice, face, fingerprint, and finger vein [2], [3]. Several security methods require touching sensors such as fingerprints and finger veins, which may increase the chances of spreading diseases such as COVID-19 [4].…”