Attack and release times must be set by ear because values in milliseconds are unreliable across compressor models

Attack and release controls are calibrated in milliseconds, but this number should not be taken literally. There is no agreed convention for how to measure or define these times, and compressor plugins implement them differently — some use logarithmic curves, some linear; some are program-dependent (reacting differently depending on the signal). This means a 10ms attack in compressor A does not produce the same sound as 10ms in compressor B. The practical consequence is that attack and release must always be set by ear, listening for specific artifacts: a fast attack catching or cutting transients, a slow attack letting them through; a fast release causing pumping, a slow release sounding natural. Ears, not numbers, determine the correct setting.

Examples

Set attack to 5ms on two different compressors on the same vocal. Listen — they sound different. Adjust by ear from the default: turn attack slowly until transients are as aggressive or controlled as you want.

Assessment

Why should a 10ms attack time not be used as a transferable preset across different compressor plugins? Name two audible artifacts that indicate attack or release is set incorrectly.