If you are a learner then you are still training your ears to pick up sounds not in your own language. I was told one of the sounds most commonly misheard by foreign ears is the "t" in 自転車 which most of us hear as "d". (According to a comment & link below, many people do say "d" but either way, even as I type 自転車 now I subconsciously type a "d" and only realise when the wrong kanji come up.)
Given that none of the other posts agree with you, it sounds like these people, like me, can hear the ん but that is not to say your ears are playing tricks on you:
ん followed by "vowels" and "near-vowels"
When ん is followed by a vowel or near-vowel type of sound, which I think applies in all your examples, it is also one the more difficult sounds to say as well as hear, and the two activities are very closely related. I still struggle to pronounce 任意 passably. Another difficult word is 禁煙. After a while I worked out that there was "y"-type of sound in many of these words and by inserting a small "y", I found Japanese were more likely to recognise them in my speech.
Likewise, in the case of 税金を the を is generally accepted as "wo" in roman letters and the combination of the "n" and the "w" is giving rise to a different sound from what you expect to hear.
The linguists and native speakers on this site may have a different view about this "y" and can explain the origins better than me. I think its roots go back to the "yi" and "ye" sounds no longer recognised in standard 平仮名. But, given that the English speaking world calls the currency "Yen", and than is pretty close to what I hear when people say ３千円 or一万円, I thinks its a good "working rule".
There maybe other differences or subtle nuances as to how ん is pronounced for different words not captured by 平仮名. The "m" sound in 日本橋 comes to mind although that is not so subtle. It also worth mentioning that some say the ん sound can be closer to the "n" in "stung" rather than in "stun".
Pitch is another very obvious example of something not captured in the writing but if the language came before the writing system then it must be inevitable that that system has pragmatic approximations.
I suppose these approximations could also have a strong influence on how the language is spoken but that is outside scope of your question and the real linguists could tell us more.