There is not perfect tool. You don't even get perfect results if you do it by hand ;-) (see e.g., https://en.wikipedia.org/wiki/Part-of-speech_tagging#Issues) The error rate depends on the domain, the genre, maybe even orthographical errors in the texts you are tagging. And you should ask yourself if you really need a tagger with a lower error rate than the Stanford tool your are using. Some problems are structural (see links above) and occure on a regular basis. Others are occuring in single cases. If your are using a large enough corpus, these error should not be a big problem.

You should just try out different tools (what language are you tagging? English?). There are many out there, e.g., the TreeTagger, the mentioned Stanford NLP toolset, and at the end of the Wikipage I linked above, you find a quite large number of tools. Then you can decide what tool comes closest to what you consider the correct POS tagging.

I assume you are referring to English POS tags. In that case, I would suggest Treetager, which seems to be very popular among my colleagues.

http://www.cis.uni-muenchen.de/~schmid/tools/TreeTagger/

See the Stanford NLP packages

http://nlp.stanford.edu/software/corenlp.shtml

you can  call these packages inluding POS in JAVA, Python, and R.

For R  See http://cran.r-project.org/web/packages/openNLP/openNLP.pdf

For Python See http://www.nltk.org/book/ch05.html

the Stanford NLP packages are very good.

recommendation.

MElt is not bad (models for French, English, Spanish and Italian ), a little better than treetagger for French  https://www.rocq.inria.fr/alpage-wiki/tiki-index.php?page=MElt

Written in Python (+numpy) (+perl for embedded tokenizer)

We have used Freeling in some projects and  we are rather satisfied

http://nlp.lsi.upc.edu/freeling/

It is available for several languages: Welsh,English,Spanish,French ,Galician, Italian, Portuguese and Russian.

We have used MontyLingua (http://web.media.mit.edu/~hugo/montylingua/). Since it is opensource we have done some improvements.

I would recommend Treetagger or any other Brill-tagger type software...

this is interesting for me and my research. Thank you researchers.

Check here a little bit

http://www.statmt.org/moses/?n=Moses.ExternalTools

I recommend you Freeling for Spanish texts