Skip to main content
xlr23
xlr23
Contributor I
September 26, 2026

Feature request — let us route the mic to our own assistant (Home Assistant) instead of Alexa

Long-time Sonos owner, first post here because I'd love to see this happen.

 

My situation: My whole home runs on Home Assistant, including a custom local voice assistant that already controls my Sonos system (native integration + the new MCP).

The workaround: I built a custom Alexa Skill just so I can say "Alexa, ask Jarvis to..." — using Amazon as a laggy, cloud-based relay to reach an assistant running on my own network.

What I want is simple: Say my own wake word to my Sonos speaker and have the audio (or just the transcript) handed off to Home Assistant on my LAN, with the response played back on that same speaker.

Here's why it's feasible: The MCP is already half the job done. Sonos 27 lets third-party AIs reach into my Sonos system. This is just the reverse direction. The HA community would build everything else (wake words, pipelines) — we only need an opt-in hook for the mic stream.

 

I'd be happy to beta test anything in this direction.

Thanks for considering it.

    2 replies

    boydfields
    Lyricist II
    September 26, 2026

    This would be awesome!

      waleri
      Lyricist I
      October 13, 2026

      +1, this would honestly be the perfect solution.

      I've been a happy Sonos customer for years. The sound and the multiroom experience are the reason Sonos is in almost every room of my apartment. My home runs on KNX and Home Assistant, completely local: lights, heating, blinds, presence, all of it. Sonos is already fully integrated there.

      The only missing piece is voice. Right now I had to build my own voice satellite (a ReSpeaker mic array running ESPHome) just to talk to my local assistant, while the Sonos speaker in the same room has better microphones, echo cancellation and far better sound. That's extra hardware, extra cables and a worse experience, for something Sonos speakers could do out of the box.

      What I'd love to see:
      - An opt-in setting to send the mic stream (after a wake word) to a local endpoint like Home Assistant's Assist / Wyoming protocol
      - The answer played back on the same speaker, with ducking of the music
      - Fully local, no cloud required, which fits Sonos' own privacy approach of Sonos Voice Control

      The Home Assistant community would happily build the wake words, pipelines and integrations. Sonos would only need to provide the hook. For many of us, this would be the reason to buy more Sonos speakers instead of separate voice devices.

      Like the original poster, I'd be glad to beta test anything in this direction. Thanks for considering it!

      Terms & ConditionsAccessibility statement