MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641107188 A) filed by Mr. M Mily Maarshiyan; Dr. M. Sughasiny; and R Aravind on September 07, 2026, for System And Method For Real-Time Context-Aware Multilingual Translation Of Sign Language Into Text And Speech.
Inventors include Mr. M Mily Maarshiyan; Dr. M. Sughasiny; and R Aravind.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: [058] A real-time assistive communication system receives continuous sign-language video and extracts temporally linked features representing signing activity. Candidate sign boundaries and sign hypotheses are generated with confidence values. An ambiguity engine selectively delays commitment of a low-confidence hypothesis for a bounded context interval and re-scores competing hypotheses using additional temporal, semantic and non-manual context. Committed sign information is converted into a structured semantic pivot substantially independent of the surface grammar of a target spoken language. A multilingual language realizer generates target-language text from the semantic pivot, and a speech synthesizer generates corresponding audible output. A latency and resource orchestrator dynamically adapts frame sampling, model path, hypothesis processing, commitment delay or local/remote execution while preserving temporal state. The architecture reduces propagation of uncertain visual classifications and permits the same recognized signing to be independently realized in different spoken languages under real- time processing constraints. Accompanied Drawing [FIGS. 1-2].
Disclaimer: Curated by HT Syndication.
