For this week’s assignment I was so fascinated by Dialogflow and I saw a big opportunity to try and create a dialog with multiple pronunciations for different words in english. As a person whose English is not their native language, I always struggle speaking to voice recognition because it's never capturing the exact words I am saying. I wanted to create a basic conversation in the supermarket and use the entities with different pronunciation mistakes. I chose words that could fit supermarket chat and I tried to use words that I know that are hard to pronounce and that the computer can not understand.
Me, Red, Fruit, Aisle, Recipe, Chaos, Flour, Draught
My first mistake was to try and write all the conversation under the same intents
It was working at first
And then it didn't
I realized I need to create the conversation in different sections
Then I added to the entities different voice mistakes that can happened
When I tried to test it there was an error with actions console
It took me a while but I realized that google actions is different from Dialogflow and that i need to sign in to Google Actions separately. Once I did it it was still something wrong with url I tried to print the error
I realized that a question mark is not something possible in Dialogflow so I had to delete all those from my intents, and then like magic it worked but, after trying a few sentences instead of hearing Foot (Fruit) it catches Food.
I went and add this to the Fruit entity and then I managed to create a proper conversation
It was really interesting creating this dialog but I also felt like it was forced, like writing each intent in separate parts so I wasn't sure if this is the right way to approach pronunciation struggles. I was happy to discover that even if I say words irrelevant it still answers the right answer but what if there is much more vocabulary then how do I separate between different words and meanings.