question

James Chivers avatar image
James Chivers asked

Intent Slot values not updating with utterance syntax

Is there any way to flush what Alexa has learnt for a given slot? In my utterances, I have something like the following: IntentName good morning {alpha bravo|NameFoo} I then update the utterance syntax to something very similar, e.g.: IntentName good morning {alpha charlie|NameFoo} I speak "good morning alpha charlie", Alexa matches to IntentName, but sends me: "slots":{"NameFoo":{"name":"NameFoo","value":"alpha bravo"}} ...when I expected to receive the updated value, e.g.: "slots":{"NameFoo":{"name":"NameFoo","value":"alpha charlie"}} I've tried wiping out IntentName completely, and testing (resulting in Alexa not knowing what I'm saying now - which is correct), but when I add the updated utterance back, it still remembers 'alpha bravo' even when it states 'alpha charlie'.
alexa skills kitvoice-user interface
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.

James Chivers avatar image
James Chivers answered
Of course, now I test again, it's updated and working :) Thus, it appears there is some lag between updating the utterance syntax in the app config, and Alexa updating it's re-learning of the new slot values - maybe around 20 minutes in my case. I'm keen to know if anyone has noticed this latency? Amazon - any advice on this?
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.

rgr@amazon avatar image
rgr@amazon answered
It is possible that you may need more sample utterances for slot generalization to work efficiently for this intent. please see: https://developer.amazon.com/public/solutions/devices/echo/alexa-app-kit/docs/defining-the-spoken-language-interface-for-an-alexa-app#Recommendations%20for%20Defining%20the%20Sample%20Utterances
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.

James Chivers avatar image
James Chivers answered
Thanks rgr; I've not experienced anything like this since, so maybe there was just some lag in processing my utterances, etc. (not sure what you typically expect to be 'ready' following an utterance update). Cheers, James
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.

Fawn@Amazon avatar image
Fawn@Amazon answered
Hi James, I've talked to the modeling team and they noted that the latency will not always be the same, but it should be somewhere between a few minutes up to half an hour or so. Please let us know if you are experiencing longer lag times, and as always, we'll be working to cut that time down! Fawn
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.

James Chivers avatar image
James Chivers answered
Hi Fawn, I really appreciate your follow-up, thank you. Having a notion of what could be an approximate lag time for new/updated utterance processing is helpful to assist with our debugging efforts. Cheers, James
10 |5000

Up to 2 attachments (including images) can be used with a maximum of 512.0 KiB each and 1.0 MiB total.