Hybrid ISL Recognition with Multilingual Speech Synthesis and Translation
摘要
The paper now puts forth a new offline application for real-time recognition of the Indian Sign Language (ISL), speech synthesis, and its translation to Indian regional languages. The developed system does not need to have any internet connectivity and, hence, can be implemented even over light CNN architectures on mobiles or other embedded systems efficiently. That is, the system should be in real time: recognizing ISL gestures and transcribing and synthesizing speech. Because this application can support various regional Indian languages, it is versatile and well suited to different users. Such an application greatly enhances accessibility for the deaf and hard of hearing population. Since it works offline, such an application can be applied a lot in the resource-limited areas due to the unavailability of reliable internet connectivity and also it presents a non-expensive scalable solution in bridging communication gaps.