You will definitely need input from the pros, quite likely outside of this forum.

Superficially, it sounds to me as though you are asking for a hardware/software combo to allow for "hands-free" operation in an essentially "hostile" environment.  Judging by your description of the "environmental variables", I'd say overcoming all of the challenges will be quite a feat, even with your work's willingness to pony up the resource.

Hands-free use of voice-recognition is the "highest form" of speech recognition, IMO.  Most of us don't and can't get to this level (also for many it wouldn't be as efficient or practical, thus not as necessary).

It also depends on whether your main goal is to control hardware/software (i.e. more along the lines of issuing commands), or to dictate prose.  The former might be easier to pull off in a noisy environment, unless you have to use short commands, simply because the permutations are fewer.  The latter is challenging if the goal is doing so hands-free, even with a conducive, quiet environment.  This alone requires a committed user and considerable effort, noise-blocking aside.

You might be better off going straight "white glove" with Knowbrainer offline, and involve your workplace's accommodation specialist or equivalent.  The other resource I'd highly consider is Lindsay's PCByVoice outfit, in the UK.  He goes by the call sign of Monkey8 here.  Both he and Lunis are essentially "hands free".

Your variables seem daunting for any definitive forum recommendations.

All the best and hope a solution works out for you!