Experiments...
It doesn't really matter what you use to carry out an experiment as long as it will suffice to do the job with sufficient accuracy.
But it does matter how you construct your experiment. If you are going to be doing a turing-esque experiment I would suggest you get a human to compose a piece of music and your computer algorithms to compose one. Then get a large sample of musicians to play each piece and say whether they were both composed by a human, both composed by a computer or one each.
You can't ask them to say one is and one isn't because by pure chance they will get this right half of the time. Expert Listeners are probably better because you may get a more conclusive result, though you may want to test this by sampling non-musicians as well.
Any turing like test is really a test of a humans ablity to spot another human intelligence and is therefore only a test of how well your code mimics a human composer.
In fact thinking about it a little more, for you experiment to be a success, ie. the listener can't distinguish human composition from that of your code, They would have to get it right exactly 50% of the time. ie. they are only relying on chance to make the judgement and are guessing.
A more value-add experiment may be to show that your code can compose music outside of the rules of human music that is pleasing to the human ear.
I'm afraid that I have no idea how your code works so can't really suggest any more. But you could try things like having lots of very simple rulesets and passing a piece's worth of random notes through the different rule-sets, varying amounts and order based on practicly anything and see if this generates anything musically interesting.
Trying serveral different methods of composing algorithms and a human composer would add more depth to the experiment because you could say that one method over the other is more successful at mimicing a human, but equally you could have a method that generates pleasing music but is audibly computer generated, would that indicate a new form of composing intelligence?
It is worth pointing out that all human composers follow rule sets whilst composing, these rule sets vary across cultures but they are there all the same. Therefore is a human doing anything more than a computer following a set of rules? Having said that one of the most telling things about great composers, in any music form, is that they know when to break the rules for maximum musical effect.
I'm going to stop now because I fear the above is mostly verbal diarrohea...
I was planning on surveying a wide variety of people (musicians, non-musicians, kids, adults, computer-savvy, computer-illiterate, etc.) but it hadn't occurred to me that a truly intelligent computer would get 50% human, 50% computer responses. Makes sense though. 
