Some posters here seem to be numpties, excuse me if that is not the case but they seem to be missing the point of the question. I post as an ex recording engineer and can appreciate what you mean [ I think!! :-) ] and what you are trying to achieve
Hope my contribution helps. I have no axe to grind and hope I have understood your question correctly. You do not state what your audio is to be used for but that really needs to be considered and you will have to make your own judgement. For some uses (see below) the iPad seems to be ideal already! If I tell you what you already know, I apologise in advance.
The frequency curves already mentioned, as you state, have no relevence to playback whatsoever but may well give a clue to the low frequency handling capability!
It would be useful, as has been pointed out, to have full specifications of response (not just the frequency range) for the units and any other useful tips for those developers who wish to produce the best quality audio output. That is not necessarily to reproduce the full audio range and as will already be evident to some listeners the ipad like other instruments can get away with a low output level in the base region without too significant a loss. Yes, it's not ideal as a playback device (you will alraedy know that!) but it is good quality for a portable device and incidentally well matched ( at least sometimes) for those who have cochlear implants as it stands, so I have been informed. Excellent for speech clarity but poor for music in the latter circumstance.
It would be hoped that the designers at Apple have made the unit respond best when fed with a standard audio signal i.e. a flat response with no compression etc. in the original waveform that sounds great when recorded/played on reference/studio quality equipment. Who knows?
In the absence of that information, and how well the dsp processes the sounds such as bit rate handling etc. it will be, without a lot of expense and time with measurement equipment for the sound output, a matter simply and for ease, trial and error. Trying to 'measure the characteristics' othewise is not easy, though not impossible, to get an estimation.
Harking back to the 'good old bad old days' heavy compression and pre-equalisation were used. It is worth experimenting to get empirical results with tracks pre recorded as flat, and some base boost and separately treble boost to see what the handling capabilities are without harmonic or other processing distortions occur on payback on the iPad. I suggest a few steps of 5dB boost at 200Hz initially but heavy bass cut below that frequency initially. Then also the same at about 10KHz. These will determine the maximum you could apply. All this done when playing back at high levels. It might be useful to use your intended sounds and also a range of others such as full frequency response music and speech as what can sound OK for on type (artificially enhanced) can sound awful with others. Once you hace the max capability established then chose pre equalisation within those parameters that will sound best and balanced.
As for compression, it does give the impression of greater loudness, but at best it is avoided as the loss of dynamic range becomes tiring for the listener. The digital reproduction and processing does not have the issues associated with older noisy recording/playback systems so ought not to be used to give a better s/n ration impression. However, as we will not be in an ideal world, it may be necessary. Certainly a bit of limiter use on the recording would be good to avoid digital overload. If the use of the sound you are producing is for within a noisy (or non quiete domestic) environment, then compression would improve the 'sound to background noise' ratio and therefore clarity. Much the same technique as was very popularly used (and still is in many cases) on some music for domestic playback on cheap equipment. However the iPad is quite good quality and clarity, so go easy on all these corrections.
If headphones are used the variety is too great to take into account so another reason to not overly apply any pre-corrections as if they all add up the result could be poor.
Hope this helps. Without definitive information (and I have not found any) it's probably best to just use your ears!