I have been waiting far too long for the subvocalization devices in SF (science/speculative fiction) to exist and become widespread. With current advances in AI

I have been waiting far too long for the subvocalization devices in SF (science/speculative fiction) to exist and become widespread. With current advances in AI/ML voice recognition I feel like we are about ready to try it, but I haven't seen any attempts yet. Anyone know if there is a project or startup working on this?