Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Several problems with voice #546

Open
auschoi96 opened this issue Sep 8, 2023 · 3 comments
Open

Several problems with voice #546

auschoi96 opened this issue Sep 8, 2023 · 3 comments

Comments

@auschoi96
Copy link

Hello all,

I'm facing several different problems with the text to speech response on my Lex V2 bot.

  1. The bot cannot read SSML and Custom Payload. I saw the previous issue about custom payload not being supported but SSML also does not work. Plain Text works fine. Is SSML also not supported?
  2. Duplicative buttons are appearing when I use voice. I have attached a screenshot. This does not happen if I converse with the bot through text.
    image
  3. Responses to the user is also given in plain text. Any formatting I set up in the custom payload is shown as is. For example, when using markdown, the text that should be hyperlinked is shown as [text](your link here). HTML/XML removes all links and just provides plain text and the bot does not read the response.

Is this a configuration issue on my end or something else?

Thanks all!

@atjohns
Copy link
Contributor

atjohns commented Sep 9, 2023

Hi,
Can you provide what the payload looks like and also make sure that you are supporting markup in the config.json file. For Markdown support the AllowSuperDangerousHTMLInMessage needs to be true.

Are you using QnA bot as well or just the Web UI - just curious in case you have some sort of custom hook that is providing that 'Did I answer your question' follow-up

@auschoi96
Copy link
Author

Hey @atjohns,

Thanks for the reply. I'm posting a screenshot where we do have that set to true
image

We are just using the Web UI. We did not use the QnA bot template. I had the slot elicit a new intent within the Lex console itself once it gives an answer. The follow up question intent is used after several different answers.

For more context, we are updating the Lex bot from a Lex V1 version to a Lex V2 version to utilize all the features that came to the Lex Console last year. This bot was made prior to those console updates in August. Previously, the voice responses formatted and responded everything correctly but that was all handled by a lambda function.

@israelias
Copy link
Contributor

@CAholder We had a similar issue when Eliciting Slot together with Eliciting a Response via built-ins on ResponseBots (with or without Document Chaining -- which you may not have had without QnA bot template). Nested buttons in a custom Slot via Console tended to produce similar or no result as well, as this seems to have been a v1 responseCard world. Version 0.19.7 addressed a v1/v2 bot init in April, and since then, we've simply followed QnABot Content Designer rules using the examples provided and have been able to avoid the dupe verification response on Fulfillment. Lex Rebuild post Slot edit was always reliable.

...In case that's at all helpful.

Adding a couple lex-web-ui-loader-config.json props we default set to true for link/text/speech co-play that are not immediately flag-shipped in the docs while i'm at it

{
  "ui": {
    "convertUrlToLinksInBotMessages": true,
     ...config.ui,
  },
  "recorder": {
    "enable": true,
    "preset": "speech_recognition"
    ...config.recorder,
  },
 ...config
}

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants