Capture DTMF input from user

There are times where user can provide information via DTMF (phone keypad presses) instead of voice. For example, when entering a PIN number, it might be easier for user to press keypad instead of speaking it aloud in public.

Currently DTMF input from user is captured and is taken into account when the agent is generating responses by default. So to set it up, you just need to prompt the agent to ask for the information via DTMF. Note that you need to ask user to say something to indicate they are done with the DTMF input.

Here’s an example prompt:

Please enter your PIN number, and let me know when you're done via speech.

And the transcript would look like this: