Limit audio without losing quality

I would also expiriment with a compressor. perhaps 2:1 with a threshould of -10 ish, with a bit of makeup gain.

Actually, limiting is the way you do this, but you have to be very careful where you set the threshold value, and there are steps you have to go though first to get any sort of an acceptable result.

So it's the Dynamics Processing module you have to use, and forget about the presets. The first thing you have to do is to normalize your signal so that the highest peak hits 0dB. This is important, because everything in the Dynamics Processor is referenced to this level. Having done that, you have to establish the point at which the slope of the graph tilts over, and how much to tilt it. This is where you have to use your ears. The idea is that it's only the peaks that you are limiting, not the main body of the audio. This is the only way to achieve the result you want - the peaks don't add to the clarity, but they are stopping you from getting all the level you're entitled to from the speaker, without distortion. Some people say that the graph should be tipped over so that it's flat - i.e. no signal above the threshold level gets through without being limited to that threshold value, but that tends to be more audible as an effect. Better to set the threshold slightly lower and have a slight tip-up on the top of the graph. Once again, it's your ears that have to judge this. When you've got an acceptable result you will be able to re-normalize the signal again, to take advantage of the headroom you've gained.

I'm not going to deny that it takes a bit of practice and experience to get a good result with this but when you have, it's pretty easy to implement. The important thing is that you aren't messing with the main body of the speech, only the peaks.