how can i improve the quality of my waves

The main formats of Pro audio are AIFF (fully standardized), Sound Designer II (advanced but proprietary), and WAVE (not standardized, but popular as heck). These three formats will show up in all of the high end programs because they allow almost any sampling rate, multiple channels, almost any bit resolution, and the allow for extra data. I suggest the WAVE format if you need a portable format, but the AIFF format if you need a less glitchy, more stable format. Since PC's can't read extra mac resource data, you may be forced to rely upon WAVEs. The only important part of an AIFF is the part that PCs can read, but too many programs go ahead and add the extra resource data. This means that if you need to transfer your AIFF's to a PC, you'll have to delete their resource forks. Sometimes you can do this by simply compressing them as ZIP or HQX files and decompressing, depending on your software. If you don't do this, you might end up with AIFFs with headers that are too big to be read correctly by dim-witted PC software. Since WAVEs are inherently PC format, mac's rarely contaminate them with extra resource data. By WAVE's I mean RIFF .WAV files. Some programs like SoundForge add tons and tons of extra non-standard junk to WAVEs. Don't be surprised if you end up finding your email address, name, operating system, date, and software logged in every WAVE file you create on a PC. All this extra data is why sometimes other programs can't read your WAVES.

OK, in terms of quality, MP-Anything is bad. so is WM-Anything. MPEG is data compression, and this means lower quality. If you want high quality, use as high a sampling rate as possible, and as high a bit resolution as possible.

24 bits at 96kHz stereo is gonna be what you'll find these days most easily supported as a pro format in AIFF, WAVE, and SD2, as well as hardware.

Some programs such as CoolEdit and WaveLab and DDClip and AnalogX AutoTune will allow 32bit WAVEs, (so do MIDIMAN/M-Audio hardwares), but they aren't always compatible, and some aren't even truly 32 bits. Anyway, 24 bits is pretty good if you use them carefully. 96kHz is also pretty good, especially if you have to use digital EQ. I have found that I can hear the difference between 48kHz 16bit songs and 32bit 96kHz songs coming out of my computer using a 32bit 96kHz sound card and an optimized reciever and speakers.

I have experimented with dithering too. I can barely notice the difference, and sometimes not at all, but dithering can help. However, I have read and confirmed with some gear that dithering should only be done when you have to downsample to 16 bits and can't stay at a higher bit res. Also, dithering does add noise, so you should only do it once if you can (right before burning a CD).
UV22 (Apogee style 22kHz) seems to be the best sounding dithering from what I hear in gossip. However, be aware that UV22 or Apogee dithering might become noisey if you use those dithered audio files or CD's for pitch-transposed sampling.

Normalization should be avoided too, because it eats up your dynamic headroom which you could otherwise use for strategic EQ which affects clarity better than normalizing which is too vague (broadband, and not specific to audible frequency content).

Knowing that each bit represents about 6 dB, you can budget your EQ and amplitude changes and mixing. at 24 bits, you know that you can beef up a fully premastered audio clip by 48 dB (24-16=8... 8x6dB = 48dB).
48 dB is great headroom for some really nice radical EQ boosts without any clipping distortion. After you've done your +48db of EQ, you don't need to (AND SHOULDNT!) normalize.

Well, that's enough beans to spill for now.

By the way, if you use a higher sampling rate for digitally EQ'd files, then the distortion caused by the EQ will be reduced and spread out. Likewise for the bit resolution. Even if you can't upsample, some programs let you "force" the sample rate to be something of your choice (header edit, document info, set sample rate, etc). This is useful because you can achieve higher quality processing (shifted slightly in frequency) and then just switch back before saving the final results. I have done this many times. It can make reverbs sound brighter too (96kHz set to 48kHz, processed, then reset to 96kHz).

Well I hope I've overwhelmed you with gobs of junk to consider.