US-German Research Proposal: ADaptive low-latency SPEEch Decoding and synthesis using intracranial signals (ADSPEED)