when I say one hundred ninety five thousand, the speech to text decodes it as 100 90 5000 if I say it slowly, and if I say it fast enough it understands it 195000 (which is the correct number)
It seems to me that it doesn't put the numbers together and always give me 195000?
Solved! Go to Solution.
ok. thanks. The reason is that we don't have control over how people will say the numbers. We have used generative AI and it tested ok.
As a conversation design tip, I would suggest telling the user how to say the number. I would suggest telling the user to say each number one by one.
You can see some examples in the pre-built Dialogflow CX components: https://cloud.google.com/dialogflow/cx/docs/concept/prebuilt-component/numeric-collection
Best,
Xavi
thanks. it is a good solution and I am doing it. Sometimes the customer says each number: 2 3 0 0 sometimes the whole number: two thousand three hundred or twenty three hundred. our requirement is to understand all. Is there a way to distinguish between these?
yeah you can do that, if you explore this pre-built component you will see how you can achieve that: https://cloud.google.com/dialogflow/cx/docs/concept/prebuilt-component/numeric-collection
is there a way to teach the system to combine these numbers using the generative AI.
for example if the person say 100 90 5000, it meant 195000 and if the person says 2400 30 it means 2430. the system gives the list of numbers but it can't combine in the above cases
thanks
To be honest. that is not a good approach, if you see the example I shared with you, you will see how to implement this solution properly
ok. thanks. The reason is that we don't have control over how people will say the numbers. We have used generative AI and it tested ok.
Hi bayfaro, how did you solved this with generative ai?
User | Count |
---|---|
2 | |
2 | |
1 | |
1 | |
1 |