In the current setup, one can use parametrised data in intents and intent_script.
However, along the docs this is limited to pre-defined values (via lists) and reduces the flexibility as one has a to create a (large) list and to continue to be ‘exact’ in the sentence
I would like to see that strings, properly identified, can be reused.
example:
‘add 5 to my account abc’ where 5 would be added to known account (entity) ABC, without the need to define 1,2,3,4,…etc.
‘send hello to mama’ where hello would be sent to the notify/mama entity
Proposal is to have this string identified based on its position in the sentence and able to pass on to script/response/etc.
The issue here is to have the intent machine figuring out what “unknown” word is.
With written text this might be somewhat easy, but once it comes to spoken ones then it becomes hard.
Think of it as translations.
The assist order is ݾ ݞ ݿ ࢥݾ
you know from the lists that ࢥݾ is “send”, ݞ is “to” and ݾ is mom, but what does ݿ mean?
Can you just say “send ݿ to mom”?
If it was done by playing it as spoken text again, then the sound for that word could just be added in, but if it was a text message, then it would no sense to the receiver.
Translating unknown words requires a lot more processing power than translating known words.
Known words can be done by a Raspberry 4 at a reasonably speed.
Unknown words require supercomputers of Google level to do it at reasonably speeds.
I agree on your view but here I want to suggest to accept ‘anything’ based on location in a sentence the user can then act on that…i.e. it leaves the user to decide and meant as added functionality, not replacing
And yes…for speaking it may become complex, but that is (again) the choice that the user makes
Assists is meant to be the base for the voice assistant in HA, so making a feature that only works with written text assists does not really make sense.
Well, Rhasspy did dabble a bit with it, so it is definitely on the developers mind.
I think the technology is too little matured at the moment, but when voice assistants take off, then the features will come too and the technology, both software and hardware will follow suite.