Potential issue/bug with repeating response values in the Assistant

I’ve been testing a very simple version of the assistant and something I’ve noticed is that if I keep giving the same query, it responds with a response to that query plus all the previous responses to previous versions of the same query. I’m not sure if this is an issue on Thunkable’s end, or DialogFlow’s end, or if there’s something I don’t understand here.

Here’s what I did:

  1. Made a DialogFlow account, copied the “Easter Eggs” prebuilt agent (one of the simplest ones, no API calls, just responses). Then I clicked on “default welcome intent” and added some “training phrases” like “hi” “hello” “hey” etc. There were already four responses there: hi, hello, good day, and greetings.

  2. I made an app (modeled off of this one) with the following code:

When I test it, it works like this:
I say “hello” when it’s listening, and it responds with one of the response phrases (for instance “good day”). Then the next time I give it a query, it will respond with the previous response “good day” as well as a new response. Then the third time I try it it responds with 3 responses. I tried this a bunch of times, it keeps going and going; I went until it was giving me 10 responses to one “hello” query.

Is this a bug? Am I missing something? Right now I’m stuck because I can’t seem to even get around it effectively; if I try to add a variable in local storage to only save the last “value”, then it won’t speak anything at all.

Here is a sharable version of the app described above:
It is missing the Client Access token though, so you will have to supply your own.

I would very much appreciate any insight anyone has into this problem.

On testing it a little more, I’ve noticed that this problem persists even if I change queries. So if my first query is “hello” and my second query is “what is the meaning of life?” I will get two responses, not one, for the second query.


Easter-Eggs too works well!

I looked at your example on my iOS 10.3 and 11.3. I have one answer every time in Label1 and everything works well. In the assistant I have.


The label is fine (it is just for debugging purposes); I also only get one response there. It’s the text to speech module that speaks multiple answers after one query.

I am using 11.4 but I only just upgraded and I’m pretty sure it was doing this on 11.3 too.

I also have multiple training phrases for the default Welcome intent, but I don’t think that should make any difference. I have also tried the other intents (including the default intent and my own intents) as well and they do the same thing.

I’m sorry, I did not understand. When I started the project, Speak did not work for me and I thought that there was a problem with Assistant. Now I’ll check it.

Interestingly, on my iOS 11.3 Text_to_Speech.Speak in the assistant’s block does not work at all.

Huh. It ought to work; I’m sure I was using 11.3 8 days ago when I first posted.
Have you allowed access to the app for the microphone and (maybe?) the speech recognizer? Do you have default responses for any query given? I’m assuming though that if you get a response in the label you probably have all that worked out.

I dunno how much you’ve played with this, but another thing to try is to use the “query” block instead of the “start listening” block; then you don’t have to worry about the microphone.

Also, it looks like in my current iteration of this I have a “stop listening” block before the text to speech block; not sure if that would make a difference but might be worth a shot.

Yes, I have everything on and Text_To_Speech works separately. I add it to the assistant block, overload Thunkable Live - Text_To_Speech does not work. It did not work and 8 days ago. I updated the firmware on iOS 11.4, reinstalled Thunkable Live - Text_To_Speech in the assistant’s block does not work. It works separately, but only after rebooting Thunkable Live after it was pulled from the assistant’s unit.

And in the block Assistant.Query Text_To_Speech works.

same thing happens to me also, it’s keep repeating the word and pick another variant of answer. Did you manage to solve this by the way?

I’ve got a way simpler version of your coding. My assistant answers correctly each time, but she repeats the answer 2-3 times each time I ask a question. Capture

yup that thing happened to me, so it is bugged?

Has anyone gotten this to work without repeating itself?

Going on 5 months with this issue :frowning: bump

Yes, it would be great to get some feedback from the Thunkable team on this issue, or at least acknowledge that it exists.

@Mark, is the team aware of this issue? Seems like it is holding people back from using the Assistant component.

Consider it acknowledged. And it is on our bug queue.


1 Like

Thanks for the reply, Mark.

This workaround seems to fix the issue in my app for the moment:

Thanks, Kyle!


Hmmm. I tried that workaround and couldn’t get it to work. I placed a Label on my main screen so I could see the value returned by the Assistant. My Dialogflow agent returns only 1 of my programmed responses, and I can see it in the Label, but she SPEAKS all the responses I have in my Intent, sometimes even repeating the first one a couple of times

Any headway on this feature?

Hi, I am getting the same error event if workaround is applied.
My experience:

  • if I send one query to dialogflow without speech recognition, it works
  • if speech recognition is used, response message is repeated 2 times

Is this issue going to be solved?