Although a having a high speed connection offers many advantages, an extremely effective presentation can be carried on 56k or 33.6K connection. If a presentation designed properly, even 28.8K connection can work effectively.

The trick is to space out the audio clips so that there is plenty of time for them to download. Of course, you can always use a "preload" segment before starting your presentation, but there are plenty of ways to avoid preloading, or to "mask" the preload so the viewer isn’t aware of it as "preloading" time.

One extremely effective method is to provide text for the viewer to read while the audio is downloading. If the text is fading in, or a "text crawl" is presented on screen, the viewer’s attention is on the text, giving you plenty of time to load the audio before playing.

Currently MP3 files offer the best compromise between fidelity and bandwidth. It’s a good idea to test various file "streaming rates" to get familiar with file size and fidelity results.

I experimented with a 10-second .wav file, and here were the resulting .mp3 file sizes created by Swish 2.0:

Note that there was no difference in the final file sizes between 44.1 & 22kHz. Substantial differences resulted depending on whether the original file was mono or stereo.