Related Experiment Video
Updated: Jan 9, 2026
05:48
Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
Published on: August 9, 2024
2.0K
A Kazakh language Dataset of Lip Movements for Command Recognition
Batyr Kenzheakhmetov1, Alissultan Amankos1, Beibut Amirgaliyev2
1School of Artificial Intelligence and Data Science, Astana IT University, 010001, Astana, Kazakhstan.
Scientific Data
|December 3, 2025
Summary
Researchers developed a new Kazakh lip reading corpus to improve speech recognition technology. This visual database aids in understanding linguistic and articulatory variations for better lip reading systems.
Area of Science:
- Computer Science
- Linguistics
- Speech Technology
Background:
- Lip reading systems interpret speech visually, crucial when audio is unavailable.
- Effective lip reading models require specialized corpora capturing linguistic and visual articulatory variability.
- Existing resources may not adequately represent diverse languages like Kazakh.
Purpose of the Study:
- Introduce a novel visual corpus specifically for Kazakh lip reading.
- Facilitate the development of more accurate Kazakh lip reading technologies.
- Support research in multimodal speech recognition.
Main Methods:
- Collected video recordings of 26 participants across a wide age range.
- Focused on 102 common Kazakh nouns with distinct articulatory patterns.
- Generated approximately 34,000 video clips, totaling 1.2 million frames.
Main Results:
- Created a comprehensive, annotated visual corpus for Kazakh lip reading.
- The corpus captures significant linguistic and articulatory diversity.
- The database serves as a valuable resource for technological advancement.
Conclusions:
- The new Kazakh lip reading corpus is essential for improving speech recognition.
- This resource will drive further research in visual speech processing.
- It holds promise for enhancing multimodal speech recognition applications.
