I built this for myself, but turned it into an integration. You need to run a Docker container to support the backend (image provided in repo), but this integration allows you to add a widget for speaking through media_player entities. Some entities support full audio, and this provides a low latency, live audio stream from mic to speaker; other media_player entites only support TTS (ie: alexa), so you can send audio messages which get transcribed to TTS. Its not two-way, its more of an announcement system.
Hi JoeHogan
I love your project!
Do you know which speakers supports live audio stream from mic to speaker?
I tried with sonos speaker but with no luck.
Thanks! I saw your issue on github… ive released a new version with a bunch of bug fixes and simplified setup. Pull the new docker image, change/remove env vars accordingly and update the integration… i see no reason this shouldnt work with sonos… it just creates an mp3 stream and then calls the home assistant play media api with the mp3 url and the media_player entity (sonos)
1 Like