In terms of your hypotheses, unless you have additional constraints on the parameters, you are going to to use the general form.
If means are the same then mu1 = mu2. What is the complement of this if you have no extra restrictions?
Hint: You need to think about what distributon and test-statistic you need to use (take a look at the various t-statistics for difference of means). Ask yourself whether you need paired or pooled t-tests first and then decide based on that what test statistic you used. (There are three 2-sample test statistics: paired, pooled, or un-paired/un-pooled).