An idea to cut down on bandwidth, is to have a system not disimilar to Escape from Tarkov, where you have F-keys (F1,F2,etc) (or any key really, EFT uses F-keys through for consistancy) allowing you to bind your common voice lines and a menu to pick from any unbound, or quickly bind them on the fly. its not hard to learn the system, and you get used to it very fast.
For example, in my settings, having not played for months I can still tell you F1 is to taunt your enemies, F2 is to agree, F3 is disagree, F4 is telling them to cease fire, F5 telling them 'good work', mostly used for taunts in PVP, F6 is telling them to repeat their last, I didnt catch what was said.. useful for gunfire masking the voices, or also taunting them if you want to be mean.
The voices are all consistant with difference voice types and such, where voice1 may say, 'hands up!' and another says 'freeze!', but they all match the theme of the game. In the case of tarkov, gruff male voice of a PMC that has been through war. In the case of starbase, we could get cool modulated voices not tooo disimilar from boltcrackers, just speaking english, and other languages could be done too, since all 'cease fire' commands are the same binding. this wouldnt be hard to have a client hear it in their own language either, though the only downside to autotranslation there is that some languages have longer or shorter phrases, so english user may be fast enough to talk with his keys to hold a conversation like normal speech (this only took me about a week to get to this level), but a german recipient may not have enough delay between commands, thus overlapping.
The idea being less bandwidth used, no mic spam trolling ruining the immersion, no low quality mic's or background noise, to both save on eardrum destruction, and again, maintain immersion. VOIP isnt the only answer for dealing with other players in game, other games have shown alternate solutions exist just fine, and I personally prefer a more immersive experience of not having weird and loud background noise from someone I am trying to interact with.
As far as companies go, managing large logistics isnt hard for a faction/company to use Teamspeak. I have personally run a multiple events with several hundred people using Teamspeak3, using channel commanders and dynamic whisper lists, allowing for a chain of command to get everyone doing what needs to be done, and keeping communications clear of unnessisary clutter. The whole setup takes about 70 seconds if the commanders havnt done it before. The server setup only a minute or two, and as low as a few seconds if you just have commanders make their own subchannels with three button clicks.
TLDR; No, we dont, other options exist that can be used in place of it just as effectively, and in some cases, more effectively and I hope that said other options are also considdered before jumping into a decision.