I am developing an Azure bot, and I am intending to link it to Cortana channel. But not sure if that need the speech to text to be part of the services needed to create that bot. or Cortana client handles the communication between cortana and the bot with the text?
Responding via speech is completely under your control. The entire Cortana experience can be driven via text entry on Windows 10 and mobile app. However, your skill may not pass certification if published because of screen-less devices like the Invoke and best practice of responding with voice is triggered by voice. You can pull DeviceInfo and fail gracefully if there is no display.