Development of novel automated language classification model using pyramid pattern technique with speech signals

Akbal, Erhan; Barua, Prabal Datta; Tuncer, Türker; Doğan, Şengül; Acharya, U. Rajendra

doi:10.1007/s00521-022-07613-7

Cited by 6 publications

(3 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dataset, consisting of 7101 sound segments representing three emotional states, achieved a 93.40% classification accuracy using the proposed 1D-OLBP and NCA-based method. The research [30] describes a novel pyramid structure for feature extraction in speech-language classification, which is applied to two datasets: a new large speech dataset and the VoxForge dataset. The approach selects 1000 interesting features using neighborhood component analysis, which are subsequently classified using a quadratic support vector machine classifier.…”

Section: Back Groundmentioning

confidence: 99%

Disfluency Assessment Using Deep Super Learners

Pravin,

Elias,

Sivaraman

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Speech Disfluency detection and classification are critical in speech therapy because they aid in tracking disfluency progression and are a major tool in technology-assisted speech therapy. Existing methods for detecting disfluency in speech are heavily reliant on annotated data, which can be costly. Machine learning algorithms are rapidly gaining recognition for assessing speech fluency, reducing human error and minimizing therapy delays, making them a preferred method over manual diagnosis. Furthermore, these algorithms do not account for variable-length disfluent speech, limiting the scalability of detection systems. For the automated detection of speech impairment, a Pipelined Deep Learner-Dual Classifier (PDL-DC) is proposed. Individuals' speech fluency was evaluated in two stages: the classification of speech disfluencies and disorders into several groups such as revisions, prolongations, whole-word repetitions, wordmedial repetitions, and filled pauses. The second aspect of classification pertains to the assessment of fluency levels, wherein speakers are classified into three categories: healthy individuals, individuals with stuttering, and individuals with Specific Language Impairment (SLI). The proposed model's implementation of a pipelined design enables the dual validation of a subject's fluency. The proposed model demonstrates an average classification accuracy, precision, and recall of 97%.

show abstract

Section: Back Groundmentioning

confidence: 99%

Disfluency Assessment Using Deep Super Learners

Pravin,

Elias,

Sivaraman

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Community emotion analysis interprets the emotional state of users from visual and auditory cues such as facial expressions, tone of voice, and body language, especially on social media and other platforms [13,14]. For this purpose, different methods have been developed using artificial intelligence and machine learning techniques [15][16][17][18]. This study presents a study on sound-based community emotion recognition (SCED).…”

Section: A Backgroundmentioning

confidence: 99%

“…𝑓𝑓𝑓𝑓 − min (𝑓𝑓𝑓𝑓) max(𝑓𝑓𝑓𝑓) − min (𝑓𝑓𝑓𝑓) (15) 𝑠𝑠𝑒𝑒 = 𝜓𝜓(𝑋𝑋, 𝑙𝑙𝑜𝑜𝑒𝑒) (16) where 𝑋𝑋 represents normalized feature vector after min-max normalization; 𝑠𝑠𝑒𝑒 , qualified index vector; 𝜓𝜓(. ) , neighborhood component analysis feature selector; and 𝑙𝑙𝑜𝑜𝑒𝑒, real output.…”

Section: 𝑋𝑋 =mentioning

confidence: 99%

FF-BTP Model for Novel Sound-Based Community Emotion Detection

Yildiz,

Tanabe,

Kobayashi

et al. 2023

IEEE Access

Self Cite

View full text Add to dashboard Cite

Background and purpose: Emotion classification models have focused on subject-but not crowd-level signal inputs. We proposed sound-based community emotion recognition (SCED) as a new machine learning challenge and have developed a novel deep learning-inspired forward-forward binary ternary pattern (FF-BTP)-based feature engineering model for crowd sentiment classification. Materials and Methods: From 187 YouTube videos, we curated a new dataset comprising 919, 905 and 909 (total 2733) 3-second recordings (sampling frequency 44.1 KHz) of crowd sounds (overlapping speech, environmental noise, etc.) belonging to verified dominant negative, neutral, and positive emotion classes. Our model architecture combined BTP-based textural feature extractor, which generated three feature vectors using signum, lower and upper ternary functions, with a novel Hinton's FF algorithm-inspired handcrafted feature vector selection function that identified the most distinctive feature vector based on maximum calculated mean square error. Our FF-BTP-based model also incorporated upstream multilevel discrete wavelet transform, which enabled multilevel feature generation in the spatial and frequency domains, and downstream iterative neighborhood component analysis-based feature selection and support vector machine classifier. Results: The FF-BTP-based model attained excellent 97.22% overall three-class classification accuracy on the SCED dataset. Conclusions: Our handcrafted feature engineering model emulated the deep feature characterization of deep learning models, and attained excellent classification results at low computation cost. It can be implemented for emerging SCED-related applications. INDEX TERMS FF-BTP; sound community emotion classification; sound processing; textural feature extraction.

show abstract

What can phone attractors in RPS tell us? A study of dynamic information in speech signals for phone classification purposes

Shekofteh

2023

Applied Acoustics

View full text Add to dashboard Cite

Development of novel automated language classification model using pyramid pattern technique with speech signals

Cited by 6 publications

References 59 publications

Disfluency Assessment Using Deep Super Learners

Disfluency Assessment Using Deep Super Learners

FF-BTP Model for Novel Sound-Based Community Emotion Detection

What can phone attractors in RPS tell us? A study of dynamic information in speech signals for phone classification purposes

Contact Info

Product

Resources

About