unfortunately, the mqtt.publisher doesn’t handle binary payloads: at the moment it’s technically impossibile and there are even good reasons for that. In order to make the Zanzito component work I had to take some “shortcut” (which I’m not sure they will pass as they are as official component…)
the only other way is an external script…
As far as I understood, it’s against HA architecture… but you can easily wrap around the service with a script: you have to define a notifier for each device, but then you can call a single script that accepts the device name as parameter and passes it to the right notifier.
Thanks for your clarification. I will use the custom notifier to achieve what I want. By the way how to use photonotification? Just curious because it didn’t explain further in the manual.
It’s just like the built in text notification, it only receives binary payloads i.e. the content of a jpg picture file. To clarify further, you can send a picture:
- with the HA notifier component;
- with a script like the example one I give in the manual;
- manually with a third party mqtt client.
Announcement:
Google Play store rules have changed and are now more restrictive:
Apps that monitor or track a user’s behaviour on a device must comply with these requirements:
…
Present users with a persistent notification and unique icon that clearly identifies the app.
I’m afraid in the next release of Zanzito there won’t be the Persistent notification option anymore…
You mean no more this…?
No, sorry, I wasn’t clear enough: there won’t be the option to turn the persistent notification off anymore
oh. it’s ok then. i never turn it off anyway.
1 Like
@masterkenobi @anon35356645 I might have a solution for multiple vocal commands for a single activity. You saw how I managed negative vocal feedbacks: a text field with multiple lines, each for a single response. I could do the same for vocal commands, but cons:
- Vocal recognition would be slowed down, especially if you have many activities each with multiple vocal commands;
- Risk for overlapping commands would increase accordingly;
- You wouldn’t be able any more to dictate the vocal command, you would have to write them down (or paste). The dictation ensured that what you said was actually understood by the recognizer in the first instance and you would see instantly the transcription. That wouldn’t be possibile anymore (it would be too clumsy to implement otherwise).
What do you think?
- would it be slower than create multiple activities with different voice cmd but with same topic?
- any difference with creating multiple activities with different voice cmd but with same topic?
- actually I prefer to type down the voice cmd. Sometimes it captures the wrong word when the words have same similar pronunciation such as “2” and “to”.
mmhh not sure, to me is actually good because you see immediately what the recognition of yuour own voice command is
Mmhh I do not like it. I like very well how is now.
AT the moment what I do is very simple, I clone the activity, edit it, and put a different voice command. SO this is actually not too bad
this is not good. Speed is definetely important.
If 1, 2 , 3 are the changes/problems, I would stay how it is now
1 Like
Try create an activity with the voice command = “set Plex volume to” and you will get what I mean.
Actually mi idea was: if I dictate something and it gets a bad transcription, it’s ok: when I say it again to give a command, it will be more probably recognized, exactly because the transcription was made by the same recognizer in the first place.
1 Like
I get it. That makes sense.
1 Like
If you looked at this thread Jarvis like automation, I posted it more than one year ago. It’s funny my dream from 1 year ago is finally taking shape. Thanks to you, point no. 2 to 5 is already covered by Zanzito. I wish point no. 1 will be possible to soon.
Thank you so much for making this come true.
3 Likes
It’s funny, yes! I hadn’t see that thread before today 
I had a look at voice/face recognition… there might be something that could be done… besides, a little more security wouldn’t be bad, would it?
1 Like