Linear phase is a property of a filter where every frequency passing through it gets delayed by exactly the same amount of time. This equal delay preserves the shape of the original waveform, preventing a type of distortion called phase distortion. The concept matters most in digital audio production, speaker design, and signal processing, where filters that treat frequencies unevenly can subtly alter the character of a sound.
How Linear Phase Works
Any filter, whether it’s an EQ on a mixing console or a crossover inside a speaker, works by changing which frequencies get through and at what volume. But filters also shift frequencies in time. A standard EQ only delays the specific band of frequencies you’re adjusting, while leaving unprocessed frequencies alone. That uneven delay changes the timing relationships between different parts of the signal, which physically reshapes the waveform.
A linear phase filter avoids this by delaying all frequencies equally, so the relative timing between them stays intact. In technical terms, the phase shift the filter introduces is directly proportional to frequency. Double the frequency, double the phase shift. This proportional relationship is what “linear” refers to: if you plot phase shift against frequency, you get a straight line. The practical result is called constant group delay, meaning every frequency component exits the filter with the same time offset.
Why Waveform Shape Matters
A complex sound like a drum hit or a vocal is made up of many frequencies layered together. The specific timing between those frequencies determines the shape of the waveform, and that shape is what you hear as the character of the sound. When a non-linear phase filter shifts some frequencies more than others, the peaks and valleys of the waveform rearrange. This is phase distortion: the signal’s frequency content stays the same, but its shape changes.
For a single isolated track, phase distortion from a standard EQ is often subtle enough that most listeners won’t notice. But in situations where multiple copies of a signal need to stay aligned, even small phase shifts can cause frequencies to partially cancel each other out, thinning the sound or creating unexpected tonal changes.
How Filters Achieve Linear Phase
Linear phase filters are built using a specific digital filter architecture called FIR (Finite Impulse Response). The defining structural requirement is symmetry: the filter’s internal coefficients must be either symmetric or antisymmetric. A 2019 paper in Digital Signal Processing provided the complete mathematical proof that this symmetry condition is not just sufficient but necessary. A FIR filter has linear phase if and only if its coefficients are symmetric or antisymmetric. No other arrangement produces a perfectly linear phase response.
Traditional analog EQs and most standard digital EQs use a different architecture called IIR (Infinite Impulse Response), which produces what’s known as minimum phase behavior. These filters minimize the total amount of phase shift but can’t distribute it evenly across all frequencies. That’s a fundamental limitation of the design, not a flaw that can be corrected with better engineering.
Where Linear Phase EQ Gets Used
Linear phase EQ shines in a few specific production scenarios where phase alignment between tracks is critical.
- Multi-microphone recordings. A drum kit recorded with several microphones produces multiple tracks that share overlapping frequency content. Applying a standard EQ to one mic can shift its phase relative to the others, causing partial cancellation that changes the tone of the kit. A linear phase EQ keeps all the tracks aligned.
- Parallel processing. When you duplicate a track or bus a copy for parallel compression, both versions of the signal eventually get summed together. A standard EQ on one copy introduces phase differences that can cause comb filtering when the two are combined. Linear phase avoids this.
- Mastering. At the mastering stage, preserving the precise waveform shape of a finished mix is a priority. Linear phase EQ lets you make tonal adjustments without introducing phase artifacts into a signal that’s already been carefully balanced.
The Latency Trade-off
Linear phase processing comes with a significant cost: delay. The FIR filters that make linear phase possible need to analyze a large window of the incoming signal before they can process it. This introduces latency that ranges from around 3,000 samples to over 20,000 samples. At a standard sample rate of 44.1 kHz, that translates to somewhere between 100 milliseconds and more than half a second of delay.
For comparison, a minimum phase EQ operates with essentially zero perceptible latency. This makes linear phase EQ impractical for live monitoring or any situation where a performer needs to hear themselves in real time. In a mixing session, your DAW can compensate for the delay automatically, but CPU load also increases since FIR filters require more processing power than their minimum phase counterparts.
Pre-Ringing Artifacts
Linear phase filters introduce a unique artifact that minimum phase filters don’t: pre-ringing. Because the filter’s symmetric design spreads its processing equally before and after the main signal, a sharp transient like a snare hit can develop a faint “chirp” or swell just before the attack. With a minimum phase filter, all ringing occurs after the transient, which is less noticeable because our ears are accustomed to sounds that decay rather than build.
Listening tests at Stanford’s Center for Computer Research in Music and Acoustics confirmed that pre-ringing is audible on sharp percussive sounds. The effect is most pronounced with steep filter slopes and aggressive EQ cuts near transient-heavy material. On sustained sounds like pads or vocals, pre-ringing is typically inaudible. This is one reason experienced engineers don’t default to linear phase on every track. They reserve it for situations where phase alignment matters more than transient preservation.
Linear Phase in Speaker Crossovers
Outside of software, linear phase behavior also matters in loudspeaker design. A crossover network splits an incoming signal into frequency bands and sends them to different drivers: low frequencies to the woofer, highs to the tweeter. If the crossover introduces uneven phase shifts, the output from the two drivers won’t sum correctly at the crossover point, creating dips or peaks in the frequency response.
Butterworth-type crossover filters are popular in speaker design partly because they maintain a mostly linear phase response within each driver’s operating range. Steeper filter designs like Chebyshev filters roll off unwanted frequencies more aggressively but sacrifice phase linearity in the process, which can color the sound near the crossover frequency. Some modern active speakers use digital signal processing to implement true linear phase crossovers, eliminating the compromise entirely at the cost of added processing delay.