FVLLMONTI

Developing VNWFETs for efficient, private in-ear translation devices, bypassing server reliance.

An innovative, inclusive and sustainable society is a European priority. For many people, inclusiveness and sustainability rely on a speech-to-speech translation obtained through a lightweight in-ear Internet of Things-based device dependent on internet connectivity. However, although machine translation has significantly improved, the use of such devices is prevented by the computation-intensive and energy-consuming artificial neural networks that need server-based implementations. What’s more, data protection and privacy concerns are on the rise. Ferroelectric vertical nanowire field-effect transistors (VNWFETs) are a solution. Through the actual fabrication of VNWFETs, the EU-funded FVLLMONTI project will develop regular 3D stacked hardware layers of neural networks based on fine-grain hardware and software co-optimisation allowing the most efficient machine translation.