Service tts.google_translate_say is translating spaces to %20

That worked, all working normally for me now, thanks all

Hey guys,
113.3 fixed %20 issue, but now there is another one, which wasn’t here before.

In a couple of automations I have a few lines of text commented out (with #), now TTS doesn’t ignore those lines.

For example in my “Good morning” automation I have the following lines in TTS message:

 #Today in Sydney it will be {{states('sensor.short_summary')}}.....
Today in Sydney it will be {{states('sensor.bom_sydney_summary_0')}}......

Instead of saying just second line it says “Hash tag Today in Sydney … etc.”
Easy fix for me just to remove those commented lines, but I keep them in many other Automations as a reference in case I need to roll back or re-enable commented line.

So basically it stopped ignoring # symbol, hope it doesn’t impact scripts, scenes and other configs.

This started happening for me spontaneously yesterday on my old HA 0.92 system, with no configuration changes. The only thing that happened yesterday was that I rebooted Ubuntu as well has restarting HA a couple of times (trying to get the alexa_media component to re-connect with Alexa).
And while substituting a dash for spaces helps, there are some issues: line breaks are interpreted as spaces by TTS, so the entire string has to be on one very long line. Times with a dash like -09:23 are not read correctly by TTS, but using a comma instead of a dash seems to be ok.
Ok, hang on here - announcements made by Node Red are not suffering the same problem. Why not?? They are also using the same google_translate_say service. Maybe Node Red passes arguments to the same service packaged differently, i.e., as JSON? So the fault must be in how HA automations are passing arguments to the google_translate_say service.

update it…