In my web app I'm using hybrid speech model as per the webchat sample code.
If adapive card has a speak field & I'm comunicating with bot using microphone then the bot is not speaking when the card is coming as response.
To get rid of this I've specified the same text in place of ssml in my Bot code as shown below:

var result = JsonConvert.DeserializeObject<AdaptiveModel>(cardAttachment.Content.ToString());
if (!string.IsNullOrEmpty(result.speak))
    await turnContext.SendActivityAsync(MessageFactory.Attachment(cardAttachment, "", result.speak, "en")), cancellationToken);

Now the bot is speaking same text twice as it has generated the Webchat have generated ssml as follows while sending it to cognitive speech:

<speak version="1.0" xml:lang="en-US">
  <voice xml:lang="en-US" xml:gender="undefined" name="Microsoft Server Speech Text to Speech Voice (en-US, ZiraRUS)">
    <prosody pitch="+0%" rate="+0%" volume="+0%">
      <s>Adaptive Card Text</s>
<s>Adaptive Card Text</s>
    </prosody>
  </voice>
</speak>

So either the bot is not speaking at all or speaking twice the same text for Adaptive card.
How to get rid of this problem?