Botframework-webchat: Speech to text issue in Safari

[enriconatella]

Screenshots

Version
Using the latest version of webchat.js (referred directly from CDN)
iOS: latest version

Describe the bug
The bug is only present in iOS Safari browser.
When a user taps the microphone and the last bot message contains a choice button, the microphone is not re-enabled and is stuck in "Starting...". From that moment on, there is no way for the microphone to work.

Steps to reproduce

• Enable Speech To Text (basic recognition)
• Tap the microphone
• Allow to access the microphone
• Say something
• Wait for the bot response (with a choice button)
• The microphone is not anymore re-enabled

Expected behavior
The microphone should be re-enabled or the user should be able to re-enabled it by pressing the microphone button.

Additional context
There is no error in the console browser.

[Bug]

[corinagum]

@enriconatella, do you have inputHintExpecting enabled on the bot's activity with the choice buttons?

[enriconatella]

@corinagum yes, in fact in all other devices is working fine (Android and web browsers).

[corinagum]

Ok thanks. Let me work on a repro - we're working on the latest Web Chat release right now, but I'll get back to you asap.

[corinagum]

@enriconatella sorry I haven't gotten to this issue yet. Would you be able to provide your activity(ies) (with no personal information) as .json? This would speed up the process for me :)

[enriconatella]

@corinagum here it is the json of the last bot activity:

{
    "activities": [
        {
            "type": "message",
            "id": "6DuhF2C3ayVmIsJnyfLnK-h|0000013",
            "timestamp": "2021-04-08T05:57:22.5315975Z",
            "channelId": "directline",
            "from": {
                "id": "p0tier1bot01",
                "name": "p0tier1bot01"
            },
            "conversation": {
                "id": "6DuhF2C3ayVmIsJnyfLnK-h"
            },
            "locale": "it-IT",
            "text": " C'è qualcos'altro che posso fare per te? 🤗",
            "speak": " C'è qualcos'altro che posso fare per te? 🤗",
            "inputHint": "expectingInput",
            "suggestedActions": {
                "actions": [
                    {
                        "type": "imBack",
                        "title": "Sì, grazie!",
                        "value": "sì, grazie!"
                    },
                    {
                        "type": "imBack",
                        "title": "No, grazie!",
                        "value": "no, grazie!"
                    }
                ]
            },
            "attachments": [],
            "entities": [],
            "replyToId": "6DuhF2C3ayVmIsJnyfLnK-h|0000006",
            "DNU": false,
            "EndConversation": false,
            "ClientData": "",
            "NodeType": "",
            "Node_QuestionType": "ChoiceSuggestion",
            "MainScenarioName": "Informazioni - Carta di Credito - Costi",
            "CurrentScenario": "End Dialog",
            "StartScenario": false,
            "ActiveNodeStep": 2,
            "SessionId": "949a8027e5304e568125fc2ee54c9e68",
            "Node_ID": "00000000-0000-0000-0000-000000000000"
        }
    ],
    "watermark": "13"
}

[corinagum]

@enriconatella sorry for the delay in getting back to you.

~I was able to resolve this issue by adding the to: [] field within suggestedActions, which is required. Could you try that and see if your issue is resolved?~

Note: When I tried this I deleted everything after entities from your JSON.

[edit]

Oops, I accidentally tested on Chrome iOS, not Safari. When I tested again in Safari, I am able to reproduce the issue.

I've added this to our candidates list for investigation/fix