They do need a warrant to listen to the voice or view/listen to communications payload data in general. They also need a warrant to remotely turn on your phones camera or microphone to spy on you (the feds have done this to various targets, particularly mafia stores I have heard about). However, they do not need a warrant to track your geospatial positioning, nor do they need a warrant to gather the list of numbers/IP addresses that you communicate with (or the list of numbers / IP addresses those numbers/IP addresses communicate with, etc). Also, they are possibly able to use traffic classifiers and DPI in order to circumvent their ban on directly viewing your communications, for example they may be able to determine the content you are viewing or even words and phrases you say with only packet timing / size information, which is obtainable without a warrant. Some commercial services that sell classifiers that can look for illegal content in internet traffic at the ISP level, make the argument that if they can prove only illegal information flows can possibly be intercepted, that it is not unconstitutional to warrantlessly analyze traffic for illegal content in a dragnet fashion. I believe this has not been decided decisively by the courts, although of course it has law enforcement support. In such a case legal communications would not be identifiable, but inherently illegal communications (ie: CP) would be instantly detected at the ISP and justify getting a warrant for a full wiretap.