Amazon Alexa - How to create a shared slot

Question

How to create a common slot for Alexa skill? So that I can create my own Todo application, it recognizes free-form text.

Answer 1

Alexa has announced the List Skill API . As mentioned above, the literal slot type is no longer supported for new skills.

If you create a custom slot with multiple meanings - depending on the expected response values ​​in one word or two words - Alexa will catch the spoken words and not in the list and transfer them to your skills. The transcription of these words is the best effort and probably not perfect. Amazon stopped using the literal slot because the results of speech to text were not good enough.

How can you register a limited beta in the List Skill API, maybe this will solve your problem.

Answer 2

Now there is a way to get the original “transcript” of a statement or part of it.

There used to be a literal type built into the slot that gave you something, but Amazon stopped supporting it.

Answer 3

Have you done any tests using a slot with some common examples in the list of slots? I ask because I have a small user slot with 10 options and I can recognize with it a very wide range of content (elements that are not in the list).

Answer 4

This article “Why a user slot is a logical solution” reports

Using custom slot types for grammar, as described above, usually satisfies this desire and improves the accuracy of NLP training. If you still want everything, you can create a custom slot called "CatchAll" and the corresponding intention and statement: CatchAllIntent {CatchAll}. If you use the same training data that you would use for LITERAL, you will get the same results. People generally find that adding slightly more scenario-specific training data improves accuracy.

If you still do not get results, try setting the CatchAll values ​​to about twenty-2 word random phrases (from the random word generator - be really random). When the user says something that matches your other statements, these intentions will be sent. When it does not match any of them, it enters the CatchAll slot. If you go this route, you will lose accuracy because you are not fully using Alexas NLP, so you will need to test hard.

And by the way, the literal slot returned in January 2017, but it is not recommended.

Based on developer feedback, we will not remove the LITERAL slot type, and you can continue to transfer skills that include its functionality.

This is a link to LITERAL Dictionary Type Link