How to get MaryTTS running with Rhasspy

Warning: The MaryTTS implementation for Rhasspy has some issues. Therefore some special configuration tricks must be executed.
Warning: The only voice I managed to get working was: cmu-slt-hsmm
I used the info in synesthesiam/docker-marytts

Installation

Run a MaryTTS server

Run a docker container with MaryTTS:

docker run -it -p 59125:59125 synesthesiam/marytts:5.2

Now we can connect to MaryTTS on http://localhost:59125 .
If you start MaryTTS on another host, replace localhost by the correct ip-number/address

In Rhasspy:

• Go to the Settings page of Rhasspy
• On the tab Text to Speech select MaryTTS
• Save Settings and Restart
• Enter URL: e.g. http://localhost:59125
• Enter Locale: en_US (or another)
  Note: default will not work
• Save Settings and Restart
• Refresh Available Voices
• Choose the voice you want (Note: For me only cmu-slt-hsmm worked)
• Save Settings and Restart

And now the trick:

• Add /process? to the URL: e.g. http://localhost:59125/process?
• Save Settings and Restart
• Test the TTS: Enter some text and press the button: Speak

Explanation:
The first URL will work to retrieve the available voices from the MaryTTS server.
The second URL is used to call the TTS wav generator correctly.
There is a problem in the Rhasspy-code that results in the wrong call to the server for WAV generation.