within the context of Digital Primer project, we work a lot with speech recognition and speech synthesis
currently we have two voices: HJK (male, english with korean accent); DO (female, german)
we use open-source coqui text-to-speech (TTS) to generate voices
project acronym proposal for UdK's AI-Förderung (BMBF, 200 x 2 mega-EUR) application (2021)
domain which I acquired for the purpose of the project
project was renamed to KKIKI and failed to get funding ...
... domain remained and is now pointing to the machine with Nvidia GPU running in GRU 313
UDK.AI is ...public collaborative cache of AI-generated content / OERs
zero-funded Baustelle & Temporary Autonomous Zone
waits in certain matrix rooms (e.g. #AI-UdK:m3x.baumhaus.digital) for messages having a form "!txt2img SOMETHING"
he passes SOMETHING to udk.ai's instance of Stable Diffusion through Automatic1111 API
posts the generated image back in the matrix room
injects corresponding "knowledge unit" into udk.ai's knowledge graph