AISLAC Joint Impact Report for SL2T 1.0

Le résumé fourni par la source

Spoken language understanding technologies have been deployed for decades, enabling features such as voice input to mobile devices, autocaptioning for videos, and assistive applications such as Live Transcribe. A comparable input modality for sign language users has been unavailable. To address this gap, Google DeepMind has developed a sign-language-to-text model (SL2T 1.0) for translating from American Sign Language (ASL) into English, and is deploying it in two Android products: Gboard and Live Transcribe. This Joint Impact Report is co-authored by Google DeepMind team and representatives of the AI Sign Language Advisory Committee (AISLAC), an advisory body convened by Google that brings together leaders from Deaf advocacy organizations, academic and cultural institutions, and professional interpreters (including representatives from the National Association of the Deaf, NTID, DPAN, and the World Federation of the Deaf). Together, we detail the practical uses, technical limitations, and operational boundaries that govern this release.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.