There are two types of prompts: normal text and audio file. When normal text is used, the gateway automatically converts it to audio using text-to-speech engine. If it is audio file, the gateway simply plays the audio file.
When a variable is used, the gateway may not know what type of item is being referred to. For example, if you use a database operation to get an audio file, and save the audio file path in a variable, then you must specify the type of the variable.
For normal text items, you may also further specify the prompt format.
For example, if you have a text item “2032345″the text-to-speech will say it as “2 million ...?, but if you specify it as "phone number", then the system will say it as a telephone number; and if you specify it as "DTMF", then the system will send DTMF tones 2032345 (as if you pressed key 2, key 0, and key 3,...).