TTS queue in Node Red for Google Cloud (and other) TTS Engine

I think issue nr 1 is related to issue number 2. I guess you use cyrilic characters that leads to these multiple underscores instead of letters.
I tried a notification with cyrilic characters and i get also theses underscores (see figure).
Screenshot 2021-07-04 at 14.17.14

When you play a notificaiton it has always to be downloaded before beeing played (reason for the delay you mention to play a message) because it could not match any notification in the cache folder due to these underscores. I will have to prepare a fix for using cyrillic characters.

I guess this are settings only available for payed google tts and thus not available here

I will have to fix this, as a workaround until fix is ready you can clear the queues with

  • msg.clearPending

for details please see here https://www.npmjs.com/package/node-red-google-notify

and then add the attribute msg.important the the next voice message.

I don’t get exactly what you mean by “turns off the speaker” do you mean it restores the initial volume ? If so, yes the idea is that each notification should be able to set it’s own notification volume level and then the initial volume level of the device has to be restored. I don’t see how i could change this behaviour.

Thanks a lot!
but how could i find out (if i not to look into NR GUI) the the node freezes and i need to clear cache? So how can i get actual node state?

I mean it goes to OFF (not IDLE) state and when it goes ON again you can hear a Gong.

You are right, you have no way to know that the node freezes without looking at the NR GUI. The aim of the clear queues is not intended for unfreezing the node, this was just a work around while waiting for the fix.
As the fix is now ready with version 1.4.6 you will not need any more this workaround. :smile:
Thus all the above listed issues should be now resolved.

The aim of msg.clearPending is to be used before playing long mp3. I have a door sensor that plays mp3 musik each time it is activated. If the door sensor is reactivated before the end of the song, then the 2nd trigger would queue the mp3 to be played again. To avoid this situation each door trigger clears the queue before playing the mp3 song.

It might just be me, but the beginning of my TTS message is being cut off. Is there a work-around for this?