Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

Improving Translational Accuracy

Improving Translational Accuracy

Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...

Improving Translational Accuracy

Improving Translational Accuracy

Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...

Associative Learning

Associative Learning

Associative learning is a fundamental concept in behavioral psychology, wherein a connection is established between two stimuli or events, leading to a learned response. This process is critical in understanding how behaviors are acquired and modified. Conditioning, the mechanism through which associations are formed, can be divided into two main types: classical conditioning and operant conditioning, each elucidating different aspects of associative learning.
Classical conditioning, also known...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

GreenAid: a confidence-weighted ensemble deep learning system for real-time plant disease detection and management.

Scientific reports·2026

Same author

MultiScaleKANNet: a hybrid CNN-KAN-transformer architecture for radiographic bone-loss risk stratification from knee X-rays.

Scientific reports·2026

Same author

FedEmoNet: Privacy-preserving federated learning with TCN-Transformer fusion for cross-corpus speech emotion recognition.

PloS one·2026

Same author

A privacy-preserving cloud storage framework with hybrid encryption, homomorphic keyword search, and blockchain-based integrity verification.

Scientific reports·2026

Same author

Few-shot android malware classification with quantum-enhanced prototypical learning and drift detection.

Scientific reports·2026

Same author

Dynamic fog node placement optimization using adaptive dynamic pufferfish optimization for real-time IoT networks.

Scientific reports·2026

Same journal

Serum vitamin D level and its association with vertigo frequency and severity in Meniere disease.

Scientific reports·2026

Same journal

PFA-Net: a physics-informed feature enhancement and attention network for interpretable bearing fault diagnosis under strong noise.

Scientific reports·2026

Same journal

Circulating inflammatory, redox, and apoptosis-related alterations in drug-naive idiopathic pulmonary fibrosis: an exploratory case-control study.

Scientific reports·2026

Same journal

A baseline-oriented dynamic aggregation approach for demand-side heterogeneous controllable resources.

Scientific reports·2026

Same journal

Temporal precision and accuracy in schizophrenia: an exploratory study.

Scientific reports·2026

Same journal

Prefrontal EEG spectral and nonlinear signatures of subthreshold depression during resting state and affectively valenced picture/video viewing: a participant-level analysis.

Scientific reports·2026

See all related articles

Search research articles

Related Experiment Video

Updated: Jun 3, 2026

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Published on: August 9, 2024

KANWhisper: leveraging learnable activation functions for interpretable and efficient arabic automatic speech

Ezzaldeen Mahyoub Naji Saeed^1,2, Belal Al-Sellami³, Mohammed Tawfik⁴

¹Faculty of Applied Science, Department of Computer Science, Taiz University, Taiz, Yemen. ezzaldeen2080@gmail.com.

Scientific Reports

|June 1, 2026

Summary

This summary is machine-generated.

KANWhisper introduces Kolmogorov-Arnold Networks (KANs) for Arabic automatic speech recognition (ASR), improving accuracy and interpretability. This novel approach outperforms existing models, offering a parameter-efficient solution for complex languages.

Keywords:

Arabic speech recognition Common Voice Interpretable AI Kolmogorov-Arnold networks Learnable activation functions Transfer learning Whisper

More Related Videos

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Published on: September 27, 2024

Related Experiment Videos

Last Updated: Jun 3, 2026

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis

Published on: August 9, 2024

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Published on: September 27, 2024

Area of Science:

Artificial Intelligence
Natural Language Processing
Speech Recognition

Background:

Arabic automatic speech recognition (ASR) faces challenges from linguistic complexity and data scarcity.
Transformer models like Whisper use Multi-Layer Perceptrons (MLPs) with fixed activations, limiting expressiveness and interpretability.
Existing ASR models struggle with Arabic's unique phonological features, such as emphatic consonants.

Purpose of the Study:

To introduce Kolmogorov-Arnold Networks (KANs) as a novel architecture for Arabic ASR.
To enhance the accuracy and interpretability of ASR models by replacing MLPs with KAN layers.
To investigate the potential of KANs in capturing complex linguistic phenomena in morphologically rich languages.

Main Methods:

Replaced MLP feed-forward layers in the Whisper model's encoder and decoder with KAN layers utilizing learnable B-spline activation functions.
Conducted extensive experiments on the Common Voice Arabic dataset.
Performed phoneme-level evaluation and layer-wise representation probing to analyze model performance and learned representations.

Main Results:

KANWhisper achieved a word error rate (WER) of 8.02% and character error rate (CER) of 2.78%, outperforming standard Whisper, LoRA-adapted Whisper, wav2vec2 XLSR-53, and SeamlessM4T v2-Large.
The model used 16 million fewer parameters (228M vs. 244M) compared to the baseline Whisper model.
KANWhisper demonstrated a 33.3% relative reduction in error rates for Arabic emphatic consonant pairs and improved encoding of these distinctions.

Conclusions:

Kolmogorov-Arnold Networks offer a viable and advantageous paradigm for ASR in morphologically complex languages like Arabic.
KANs enhance both recognition accuracy and model interpretability, providing insights into linguistic feature processing.
This research opens new avenues for developing parameter-efficient, accurate, and interpretable ASR systems.