What is the best open-source options to create a pipeline like ElevenLab (Speech-to-text, brain LLM and text-to-speech)

I want to create a pipeline locally hosted and we can't use a outsource provider due to regulations. There are two ideas in my head. 1- Create a locally hosted pipeline, if so what are the best way to overcome this? 2- Find a way around to use ElevenLab (maybe redact sensitive data or some other tec