Full-duplex / native speech-to-speech support?

Is native full-duplex/speech-to-speech on the roadmap? If so, rough timeline?