I’d like to propose some upgrades to the voice pipeline and would also like to know how and where changes would need to be made in to make them a reality. I’m not really sure where to start as I currently am basing this off of my experience using MicroWakeWord on and ESP32 satellite, then communicating with the default HA conversation agent, with audio coming back out the ESP32 satellite.
- Wake words and commands in a single sentence without a pause after the wake word.
- Being able to interrupt a satellite while it’s talking, to change the last command, or to issue a new/more commands.
- Keeping the mic feed from a satellite open for a programmable amount of time while and after it’s responding to a command for a more conversational feeling.
4?. Not sure if this is already implemented, but being able to have multiple satellites in an area and if both pick up the wake word/command, picking whichever one the system determines to be closer to the person to use as the mic/speaker for interaction.