What are the challenges of building a self-hosted voice infrastructure?

In a recent project, I faced numerous challenges while developing a self-hosted voice application. Key issues included integrating with multiple communication providers, managing API complexities, and ensuring reliable call recordings. I built a proof of concept that took control of SIP signaling and media handling, simplifying the process and improving visibility into phone calls. I'm now open-sourcing this project, Comcent CE, and would love to hear your thoughts on the challenges of building such infrastructure and any experiences you might have had.