Emotion Recognition from Mizo Speech: A Signal Processing Approach

Mangalam, Aditya Raj; Singh, Surender; Lalremtluanga, C.; Prabhash, Kumar; Das, Ranjita; Basu, Joyanta; Chatterjee, Saibal

doi:10.1109/icdcece53908.2022.9793078

Cited by 2 publications

(1 citation statement)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cepstral analysis, coupled with a frequency scale developed by Davis and Mermelstein in the 1980s, emerged as a potent tool for speech signal processing and natural language recognition [11]. Mel Frequency Cepstral Coefficients (MFCC), which extracts features from signals in the frequency domain, help reduce dimensionality and aid in anomaly detection in brain activity [12,13]. MFCC offers benefits such as noise resistance and pattern identification in the frequency domain [14], and it has been successful in speech recognition and natural language processing [15][16][17].…”

Section: Introductionmentioning

confidence: 99%

Autism Classification and Identification of Significant Brain Lobe Using Cepstral Coefficients

Shanmugam,

Radhakrishnan

2024

View full text Add to dashboard Cite

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by restricted, repetitive behaviors and impaired social interaction. Currently, the identification of individuals with ASD largely relies on subjective assessments, presenting a challenge for researchers to distinguish between Typically Developing (TD) children and those with ASD. This study analyzes EEG data from 10 children with ASD and 10 TD children in response to an audio-video stimulus. Two separate analyses were performed on EEG frequency bands within the range of 0-70 Hz and specific frequency bands of 8-30 Hz, aiming to identify the brain lobe region that yields the most significant discrimination between ASD and TD. Parameters such as Linear Frequency Cepstral Coefficients (LFCC), Cepstral energy, signal energy, delta, and delta-delta derivatives were utilized for the analysis. The study deployed classification techniques including K-Nearest Neighbors (KNN), Multi-Layer Perceptron (MLP), Decision Tree, Support Vector Machine (SVM), Bagging KNN, and Random Forest (RF). The results indicated that KNN surpassed all other classification models for frequency bands within the range of 0-70Hz, achieving a discriminating accuracy of 98.3% for ASD and TD in the central lobe region (C3, C4, Cz). However, KNN did not yield a significant level of accuracy when applied to a specific frequency band; it was improved by employing Bagging KNN, reaching 93.8% in the central lobe region (C3, C4, Cz). The electrode combination in the central lobe (C3, C4, and Cz) demonstrated superior discrimination between TD and ASD compared to other brain lobes.

show abstract