I have developed an Alexa skill, but its speech-to-text transformation is not precise enough for my backend. I have intents like "Whats the {playername}'s performance". These player names are real nicknames, so they are kind of tricky to be analysed. Is there a possibility to add like specific words to the skill to enhance precision, so it first tries to match vocals to these set of words?
Up to now I didn't find any information in any documentation. I'm generally new to working with Alexa skills.
So instead of "How many goals did the jeep king score?", there should be precious word matching, so ideally I get a result like "How many goals did TGKing score?"
I hope it's understandable what I am looking for.
Alexa custom slot types might be an appropriate solution to adopt.
The custom slot types is an extension of the slot types already built into the Alexa Skills Kit. If you're using English (US), you might find that one of the already defined slot types are enough, i.e. AMAZON.Athlete, AMAZON.Person, or AMAZON.Professional.
Use this in your sample utterances by, for example, specifying nameOfPerson as the "intent slot" with AMAZON.Athlete as the "slot type". So in the example you provided, the sample utterance would be "What's the {nameOfPerson}'s performance"
If the built-in types aren't specific enough, you can use the custom slot types to define your own type, which you could call for example player. When creating the slot type player, you will specify every player you want to include and you can also enter any synonyms (e.g. nicknames) which should match to the player.
Have a look at the documentation - which provides an insight into how to edit and upload custom slot type values in bulk, as well as how to extend the built-in slot types with additional values.