question

salazarpresentys avatar image
salazarpresentys asked

how to avoid ?

Hi,

why does <audio only response> message appears in Alexa Developer Console Test Simulator? It seems like the input is so long that Alexa is not able to process it. But is only an hypothesis, because I cannot see any logs or json output.

Could somebody help?

Regards,

Pedro.

alexa skills kitalexalambdanode_js
3 comments
10 |5000

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

CharlesF avatar image CharlesF commented ·

There is a limit to the size of the response of 2000 characters or so. I don't remember the exact amount but it's pretty large. I just did a quick test and when you go over that limit Alexa comes back with "There was a problem with the requested skill's response" so that is not what your problem is.

If you reply back here with some of your code we can try to figure out what is going wrong for you.

0 Likes 0 ·
salazarpresentys avatar image salazarpresentys CharlesF commented ·

Hi Charles!, thanks for response.

I am not able to share any code for reasons about business confidentiality and my Handler is so deep to try here.

I can say my skill works fine but if I say, for example: "Alexa, search for {name}" where {name} could be as long as: "This is an example about how is my input and my input could have decimal like 4,53 ml or 0,348 mg, for example."

Hope this would be able enough for understanding how my inputs are.

Regards,

Pedro.

0 Likes 0 ·
CharlesF avatar image CharlesF salazarpresentys commented ·

Based on what you are seeing, it seems to me that you are not forming a proper response. Without seeing any of your code it is hard to guess why that might be. Here are a few things that I do know.

  1. The issue is not that your response is too long. If it was you would get a response back indicating "There was a problem with the requested skill's response".
  2. You are indicating that you have a "name" slot that contains lengthy sentences. I have never done that and I think this is not likely to work well. Slots in the interaction model are intended to be atomic bits of language like names, places, colors, numbers etc. From what you are describing, it looks like a better interaction model would be to have a "quantity" slot which expects a number and a "units" slot which expects your valid units. I would recommend pre-built slots like these: https://developer.amazon.com/docs/custom-skills/slot-type-reference.html#list-types
  3. You would get this result if the response was not formed correctly. Try dumping the speech object you are sending back to Alexa to your Cloudwatch. Run that output through the Voice&Tone simulator in the Test tab of the Developer Console and see what happens.

Good luck!

1 Like 1 ·

1 Answer

Rokas avatar image
Rokas answered

Here, in this blog post, you can read more about response types that alexa can return. One of them is audio only response, which could mean that response you're sending is not understood by alexa.

10 |5000

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