Compression

As mentioned in Chapter 2, "Preparing Yourself," lossless (as opposed to lossy) audio compression smoothes the audio's dynamic range. Although the concept of compression is similar to normalization, the method and result are rather different. While normalization changes the audio level consistently by the same amount, compression makes softer and louder parts closer to the same level.

When encoding to low bit rates, use compression for audio that has a wide dynamic range to allow the encoder to do a more efficient job. Classical and acoustic music benefit from a healthy dose of compression prior to encoding. Sadly, doing this results in a loss of sonic subtlety, but most streaming audio (especially low bit rate) loses this anyway.

NOTE

The results you get from identical settings will vary between hardware and software compressors due to variations in dB levels and the different ways that analog and digital domains handle clipping. Read your documentation, refer to some of the resources in the Appendix, and experiment to make sure you're getting the most out of your gear.

Adjusting different compression parameters can alter the sound of your audio subtly, dramatically, or anywhere in between. Compressors are available as software, either independently or as part of a bundle or plug-in for Macintosh and Windows, and as standalone hardware. You can acquire serviceable compression tools for less than $100, but you can also easily spend $5,000 on a single channel of classic, mid-sixties, military-spec, all tube compression hardware with a sticky VU meter. Many hardware compressors are available, new and used. A Web search on compressors and compressor plug-ins quickly reveals a large number of products.

Typically, using lossless audio compression to prepare your audio for encoding doesn't require a heavy hand. A little bit of dynamic range smoothing should be sufficient.

Following are a few key terms and what they mean:

• Attack Indicates how quickly the compressor acts on the input signal. A lower number is a faster attack, measured in milliseconds.

• Release Indicates how quickly the compressor stops acting on the input signal after the signal has gone below your threshold setting. A lower number is a quicker release, measured in milliseconds.

• Output Indicates how much to increase the volume level after the signal has been compressed. This is to make up for any lowering of the overall volume level during compression. This is measured in dB, and a higher number is a louder volume increase.

• Threshold Indicates the audio level where compression starts. This is the point at which the compressor is told to start compressing. Threshold settings are usually marked in dB, measured from a 0dB point (the current setting's point of distortion). The compressor compresses all sound above that threshold.

• Ratio Indicates to the compressor how much to bring down the volume level. A setting of 1:1 has no compression. A higher setting (such as 4:1) brings levels down more. Infinity:1 brings the sound level down to exactly match your threshold setting.

Following are three basic examples for using multipurpose dynamics compression to optimize source audio that will be streamed at low bit rates. These templates are merely starting places. As you experiment with different settings for different kinds of audio, you'll become comfortable with these terms, the processes they represent, and the results achieved by using them.

NOTE

A good combination hardware tool is a compressor/peak limiter. It enables you to compress the audio and also prevents peaks in the signal. Avoiding peaks is good in any audio scenario, not just streaming audio.

• General purpose (mild compression) Set Threshold to 10dB. Set Ratio to 4:1. Set Attack Time to 100ms. Set Release Time to 100ms. Adjust the Input Level to get approximately 3dB of compression. Adjust Output Level to be about 0dB.

• Voice only (heavy compression) Set Threshold to 30dB. Set Ratio to 40:1. Set Attack Time to 10ms. Set Release Time to 100ms. Adjust the Input Level to get approximately 7dB of compression. Adjust Output Level to be about 0dB.

• Music Set Threshold to 0dB. (This threshold setting is for music with a narrow dynamic range. Lower the Threshold for music with a wide dynamic range.) Set Ratio to 20:1. Set attack time to .5ms. Set release time to 50ms. Adjust the Input Level to get approximately 5dB of compression. Adjust Output Level to be about 0dB.

FezGuys' Tip

If your listening audience tunes in from a workplace environment with a lot of ambient noise using small speakers, you'll want to use more compression to prevent quieter passages from getting lost.

Professionally mastered audio sources (such as modern pop music CDs) usually don't require too much compression optimization prior to encoding; use equalization instead (see the next section). However, compression is especially important when streaming a live signal (such as a live band or a person speaking). There's no right or wrong way to compress your audio. Let your ears be your guide.