The problem could be with the limiting applied by the H2. Ideally, you want as strong a signal as possible without crossing the clipping threshold. And that's why some people have a tendency to automatically apply limiting and compression to their recordings.
While these portable recorders do an amazing job for the money, they are not built with top quality components and processors. Some audio people might describe your recording as "crunchy" or "hot," and that is sometimes a result of too much limiting or poor quality limiting. You would probably be better off recording naked, with no processing. You can always add limiting and compression to even out your sound in Audacity.
That means you might have to set the levels yourself, but that's usually not a problem with interviews. I have an H4, and it offers a choice between compression and limiting, but it also has an option for "auto gain," or AGC. (I'm pretty positive that the H2 offers the same options.) Instead of continually monitoring the input and adjusting the gain to optimize the levels, it scans the level while in record-ready mode and sets a static value based on that. This eliminates the unpleasant "pumping" that can result from constantly-shifting gains, especially overshoot from a brief loud sound. Your levels may be a little lower than desirable, but you can really clean things up in post processing. Normalize your recording. If you have some noise or hiss, Audacity has a pretty good noise reduction plug-in. Using the built in compression or limiting in the H4 and H2 can really amplify any background noise, as it is, and the constantly changing gain can make the noise more difficult to remove. If you set your own levels or use AGC, any noise will remain at the same level and is easier to reduce or remove.
If the volume of your subject is a little erratic (gets louder and softer), it's better to squeeze your signal a bit with compression or limiting in Audacity because you have control over the amount applied.
So, anyway, try making a recording in the WAV mode and setting your own levels, or use AGC. Then do any other processing in Audacity, or whatever audio editor you might use. Avoid recording in MP3, then exporting in MP3 -- it's akin to shooting in JPEG, then re-saving in JPEG after editing -- every time you recompress, you lose more information and more quality.
Also, if you are creating audio files to import into Soundslides, I would highly recommend downloading the LAME component for Soundslides. Export your audio from Audacity as a WAVE or AIFF file so you always have a full quality version. Then, when you import into Soundslides with the LAME component, it automatically creates a good quality MP3 version at an appropriate bit rate.
(Honestly, your slightly crunchy recording isn't nearly as bad as some of the jingly jangly, artifact-ridden audio that I've heard as a result of low bit rates. Low bit rates save band width and storage space, but they sound so bad that they devalue and interfere with the telling of some potentially good stories.)
Hope that helps. Write back and let us know.