Streaming audio from python to Rhasspy


I am trying to bridge audio from ODAS to Rhasspy (2.5.7). I am streaming the audio from odas as raw and converting the chunks to wav (16bit 16kHz) with valid wav header using python.

Now I am wondering how to stream the chunks to Rhasspy for wakeword and intent recognition? I tried by publishing the chunks to MQTT (hermes/audioServer/default/audioFrame). I can see them getting through from Node-Red but Rhasspy is not reacting, so I am obviously doing something wrong or I have misunderstood the documentation.

Any tips would be appreciated.
Thank you for taking your time