Both types can draw the same curve on the screen. They differ in what they do to timing, and that decides which one to use for a given job.
Minimum phase
This is how analogue filters behave. Boost or cut a band and the signal also shifts in phase around that band, in a fixed relationship to the amount of gain. It adds no delay and uses little CPU. Its impulse response starts immediately and rings after the main hit, which the ear is used to and tends to hide.
Linear phase
A linear-phase filter delays all frequencies equally, so the phase relationship between them is preserved. The cost is latency and a symmetrical impulse response, so the ringing also happens before the transient (pre-ringing). Narrow, steep boosts and cuts make that audible on drums and other sharp sounds. It also needs more CPU.
When to choose which
| Situation | Pick |
|---|---|
| Everyday tracking and mixing, individual sounds | Minimum phase |
| Anything where low latency matters (live, monitoring) | Minimum phase |
| Processing several mics or paths of the same source that must keep their relative phase (drum overheads, parallel chains) | Linear phase |
| Mastering with gentle, wide moves | Either, linear phase is common |
| Steep high-pass or low-pass on bass-heavy material | Be careful: both have side effects, listen |
Which one I reach for
Minimum phase is the right default for most work. Use linear phase when related signals must keep their phase relationship. Accept the latency, and keep boosts wide and moderate so pre-ringing stays below hearing.
If you are unsure, A/B at matched loudness and go with what you hear.
