one of my automation pulls news from RSS, feeds it to LLM and then let Voice PE read it out via announce service. The problem is that most of the time it does not work - I hear the preannounce sound and that’s it.
I haven’t checked it in depth but I do wonder if it has something to do with a memory limit on ESP32? Should I instruct LLM to generate report that is less than n chracters long?
It might be helpful if you share your automation.
Where does the automation get the data from the RSS feed from? Could it be that you store it in an input_text, which is limited to 255 characters?
This question will be repeated until the developers complete the development of streaming response synthesis.
Most likely your tts does not have time to prepare the text in 5 seconds.
you can test this separately by running the tts service. Or look at the logs in Esphome