Abstract

Telling humans apart from automated bots has long been a frustrating challenge for visually impaired users, who find themselves locked out of ordinary web services by CAPTCHA systems built almost entirely around vision. Distorted text, image grids, and moving puzzles offer little to someone who cannot see them clearly, or at all.

This paper presents a CAPTCHA system built around a different approach: rather than asking users to see something, it asks them to hear and then speak. A word is drawn at random from the system's database, played aloud to the user, and the user repeats it. A speech recognition engine then compares what was spoken against what was generated, completing the verification without any visual interaction. The design rests on a straightforward observation fluent, natural speech remains something humans do far more organically than machines  and the two-step process of listening then responding adds a meaningful layer of resistance against automated attacks.

The system was tested alongside currently deployed CAPTCHA solutions, with forty-seven participants spanning four categories of visual impairment: total blindness, partial blindness, night blindness, and half blindness. Results showed a 70 % rise in verification success rates and a 95 % reduction in access time compared to existing systems. Participants who previously required sighted assistance were able to complete verification independently using only their hearing and voice.

Beyond accessibility, the spoken-word database collected during testing opens a path toward future Arabic-language speech-to-text development, allowing visually impaired individuals to contribute to rather than simply be accommodated by the technical projects that shape their digital lives.

References

1. Fanelle, V., Karimi, S., Shah, A., Subramanian, B., and Das, S., Blind and Human: Exploring More Usable Audio CAPTCHA Designs, Proceedings of the Sixteenth Symposium on Usable Privacy and Security (SOUPS 2020), pp. 111–125, 2020. [Online]. Available: https://www.usenix.org/conference/soups2020/presentation/fanelle
2. Alnfiai, M. M., A Novel Design of Audio CAPTCHA for Visually Impaired Users, International Journal of Communication Networks and Information Security (IJCNIS), vol. 12, no. 2, pp. 168–179, 2020.
3. Abdullah, H., Karlekar, A., Prasad, S., Rahman, M. S., Blue, L., Bauer, L. A., Bindschaedler, V., and Traynor, P., Attacks as Defenses: Designing Robust Audio CAPTCHAs Using Attacks on Automatic Speech Recognition Systems, arXiv preprint arXiv:2203.05408, 2022. [Online]. Available: https://arxiv.org/abs/2203.05408
4. Hossen, M. I., and Hei, X., aaeCAPTCHA: The Design and Implementation of Audio Adversarial CAPTCHA, arXiv preprint arXiv:2203.02735, 2022. [Online]. Available: https://arxiv.org/abs/2203.02735
5. Alqarni, M., Yu, F., Raavi, R., and Sukhai, M., Audio and Facial Recognition CAPTCHAs for Visually Impaired Users, in Encyclopedia of Computer Graphics and Games, N. Lee, Ed. Cham: Springer, 2024. https://doi.org/10.1007/978-3-031-23161-2_508
6. Mulazimoglu, E., Cakir, M. P., and Acarturk, C., The Role of Visual Features in Text-Based CAPTCHAs: An fNIRS Study for Usable Security, arXiv preprint arXiv:2311.18436, 2023. [Online]. Available: https://arxiv.org/abs/2311.18436
7. Alnfiai, M., and Alassery, F., TapCAPTCHA: Non-Visual CAPTCHA on Touchscreens for Visually Impaired People, Journal on Multimodal User Interfaces, vol. 16, pp. 385–398, 2022. https://doi.org/10.1007/s12193-022-00394-2
8. Wang, P., Gao, H., Guo, X., Yuan, Z., and Nian, J., Improving the Security of Audio CAPTCHAs With Adversarial Examples, IEEE Transactions on Dependable and Secure Computing, vol. 21, no. 2, pp. 650–667, 2024. https://doi.org/10.1109/TDSC.2023.3236367
9. Alnfiai, M. M., Evaluating the Accessibility and Usability of a Universal CAPTCHA Based on Gestures for Smartphones, Universal Access in the Information Society, vol. 19, no. 4, pp. 843–855, 2020. https://doi.org/10.1007/s10209-020-00730-x