Cancelable speaker identification system for the handicapped access to healthcare networks

Abstract One of the most promising applications of speaker recognition technology is to assist individuals with disabilities. With the rapid advancement of the Internet of Things (IoT), integrating speaker recognition into IoT devices is becoming increasingly important, particularly for secure access in healthcare settings. Biometric authentication using speech signals offers a compelling solution due to the distinctive and robust features of human voice. This paper proposes a cancelable speech encryption algorithm based on the fast Walsh–Hadamard transform (FWHT) and the RSA encryption algorithm, designed specifically for IoT network access control. The speech signal is first transformed and compressed using FWHT to reduce intelligibility in the transform domain. Security is enhanced through the combination of FWHT and the RSA encryption scheme. The effectiveness of the proposed method is evaluated using the TIMIT and Chinese Mandarin speech corpora, demonstrating promising performance. The approach achieves an equal error rate (EER) of 5.6184e-04 on the TIMIT dataset and 0.0400 on the Chinese Mandarin dataset, with an area under the ROC curve (AUC) of 99.99% for both. While these results demonstrate the potential effectiveness of the framework for securing networks through cancelable biometrics and fast processing time, the study does not include an evaluation of computational complexity and scalability in resource-constrained IoT environments. Future work will address these limitations to further assess the robustness and practical applicability of the proposed approach.