Tether has developed QVAC, an open-source AI stack that enables on-device processing of AI models, including a new brain-to-text model called BrainWhisperer. This model, developed by Tether Evo's research team, can decode neural signals directly into text with a reported word error rate of 8.7% for a single participant, while operating within 2GB of memory. The integration of BrainWhisperer into the QVAC SDK aims to provide developers with a foundation for assistive technologies, particularly for individuals with conditions that impair speech. AI
IMPACT Enables developers to build assistive technologies for individuals with speech impairments using on-device AI.
RANK_REASON This is a proof-of-concept integration of a novel AI model into an existing on-device AI stack, rather than a direct release of a frontier model by a major lab.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
