Talksign, a Nigeria and United Kingdom based Artificial Intelligence research company, has launched two AI powered models designed to enable real time translation between American Sign Language and spoken or written language as part of efforts to improve communication accessibility for people with hearing impairments.
The newly introduced models, Palm 1.0 and Echo 1.0, were unveiled on May 20 and build on the company’s earlier Talksign 1 foundation model launched in February. According to the company, Palm 1.0 translates American Sign Language into text or speech with 84.2 per cent semantic accuracy, while Echo 1.0 converts written or spoken language into photorealistic sign language video using AI generated avatars in real time with minimal delay. Talksign explained that the systems were developed to overcome limitations in earlier models that could only interpret isolated signs rather than continuous conversations and fingerspelling.
Speaking on the development, the Chief Executive Officer and co founder of Talksign, Edidiong Ekong, said Palm 1.0 represented the company’s first model ready for large scale use by Deaf communities across devices including smartphones, smart glasses, classrooms and hospitals. The company disclosed that Palm 1.0 was trained on more than 71,000 ASL samples using a transformer based architecture capable of tracking 133 anatomical landmarks on the body to interpret signing as contextual conversation rather than isolated gestures.
Talksign further stated that Echo 1.0 was trained on more than 94,000 ASL sentence pairs and could generate 30 frames per second sign language video with near real time latency. The company added that the models were developed with contributions from Deaf advisors, educators and accessibility advocates, while future versions would support additional languages including British Sign Language and Nigerian Sign Language. Full deployment of the models on desktop applications and Meta Ray Ban Smart Glasses is scheduled for August 20, 2026.