Since the sound quality of streaming audio is so often defined by its limitations it's helpful to do a little modifying to the overall tone and shape of your source audio. Don't worry, the process is pretty straightforward and, once you've tried it the first time, you'll realize it's actually kind of fun. You're changing the way your source audio sounds and you'll find you can do some odd things. Maybe you won't actually end up saving your source audio in those odd and interesting ways, but it's nice to know you can do it, just the same. Below are a couple of tools and techniques to get you started.
The Gain Chain
From the first time a sound source is recorded, and with each succeeding generation, the overall quality of your audio is defined by the gain setting. The gain is simply the volume at which the sound source is recorded onto the recording medium (tape, minidisc, DAT, and so on). Too high a gain setting creates unwanted distortion. Too low a gain setting requires the listener to turn up the volume so high that the inherent distortion of that recording medium (usually below the threshold of hearing at normal volume levels) becomes plainly audible. Both scenarios are bad juju. The key is to set the volume level as high as it will go within any given recording medium without distorting. Trust your ears here. If it sounds good, it usually is.
The human ear and brain are typically not too keen on sounds that begin and end abruptly. If you are creating short audio streams from a longer piece of sound, fade in the beginning of the segment and fade out the end. Experiment with fast and slow fades to see which makes more sense for that particular piece of audio. Your waveform editor is the application to perform these simple functions. It is also helpful to place one or two seconds of silence at the beginning of your streaming audio file. Before your online listener actually hears a sound, they'll often experience a brief awkward pause while the server accessed by their computer gets its act together. This is called buffering.
It's a good idea to add an extra second or two of silence at the end of the song as well. Sometimes formats lose the last second or two of your streaming audio file.
Compression (standard lossless audio compression) is a gentle massaging or sculpting of sound to smooth out the difference between loud and soft sections. Compression is useful for preparing audio files for encoding. Streaming formats, after being piped through online networks, tend to sound better if the pre-encoded source audio is all about the same volume. A waveform editor is the tool to use here. Audio gurus spend lifetimes understanding the nuances of perfect compression. But this is only streaming audio, so don't stress it. Most waveform editors provide a few basic settings. Try a little bit of compression on your audio file, prior to the encoding process. Let your ears be your guide.
Encoding and decoding an audio file for the streaming process is also a form of compression, called lossy compression. In lossy compression, elements of the sound are permanently thrown away to save space.
Speaking of streaming audio's ability to deliver decent sound, did we mention carving up the sound frequency spectrum? Format codecs are able to squish large raw audio files down to a tiny size by removing parts of the audio that are seen as redundant. Using your waveform editor, perform the following basic EQ changes to your source audio to help them be better sounding streaming audio files, especially when encoding at low (below 56Kbps) and medium (up to 96Kbps) bit rates.
If you do plan to encode at low to medium bit rates, especially when using the live encoding scenario, remove the lowdown lows (39Hz and below). Your audio file won't need them where it's going. While you're at it, toss out the high-end highs (18kHz and above), too. When encoding for streams below 56Kbps, experiment with boosting mid frequencies (around 2.5kHz) to compensate for the frequencies you threw away. The overall sound quality of low bit rate encoded spoken word files especially benefit from this treatment.
All this sonic manipulation won't sound very good when compared to the original source recording, but it'll translate a lot better through the encoding process as a sonically legible file. The encoder (when encoding at low bit rates) removes these frequencies anyway. Performing these EQ changes prior to the encoding process frees the computer to spend more time doing a good job encoding the frequencies that matter. You won't have to adjust much unless you're actually removing certain frequency ranges. A little boost or cut of certain frequencies goes a long way.