Just to clear some things up -
The sample rate refers to the number of times a second the computer takes a 'snapshot' or 'sample' of the original sound wave to convert it into 1s and 0s. The emphasis here is on the word conversion. As soon as we need to 'convert' something from one medium to another, we are going to inherently lose some of the original...just bear this in mind.
For example, if we sample something at 44.1kHz the computer will be taking 44,100 snapshots of the original sound a second. Conversely, for 96kHz the audio will be sampled 96,000 times a second. Therefore, the latter is offering a more accurate or 'true' representation of the original source.
The bit depth refers to the number of bits you have at your disposal to capture each sample of audio. This will directly affect the dynamic range of the recorded audio and the signal-to-noise ratio.
Redbook standard CDs (all audio CDs which you buy in the shops/online) adhere to 44.1kHz / 16-bit. It is important to remember that sample rates and bit depth are ONLY relevant for digital audio as a computer only understands 1s and 0s and NOT analogue sound waves...leave this to your speakers!
Now, as you can imagine, the higher the sample rate / bit-depth, the higher the overall strain is placed on your computer and audio interface or soundcard. This is why, I am such an advocate of 96kHz / 24-bit, as in my opinion, (and you must remember I have been working in pro audio for a while now) there is hardly any difference between 96 and 192. Or at least, to be able to hear the difference, one must spend £2,000+ on professional monitoring and acoustic treatment to be able to tell.
Also, since Windows Vista (when Microshaft re-wrote the way audio is processed by Windows by off loading it from the kernel) there has been an annoying bug whereby; irrespective of what samplerate and bitdepth one sets in the audio driver panel for his or her interface, Windows would not sync its own sample rate and bit depth to match. This results in a very noticable loss of audio fidelity and the only way I can describe the loss is the difference between say, a 112kbps MP3 and a 320kbps one.
This issue is easily rectified by matching your soundcard's sample rate and bit depth to Windows (I would personally always recommend 96kHz / 24-bit) to prevent any audio quality degredation by Windows.
Furthermore, by setting 96kHz / 24-bit you are not necessarily matching your source material (as someone mentioned earlier) but this does not matter. 44.1, 48, 88.2 will all work fine with 96kHz being set as your master sample rate but will prevent Windows from degrading your audio. You spent the money on the soundcard - why shoot yourself in the foot by using the lower sample rates?
