When measuring a room’s impulse response, engineers use a sinusoidal sweep (e.g., a logarithmic chirp). The recorded response is convolved with the inverse allpass filter of the original sweep. The resulting relies entirely on the known allpassphase of the sweep signal to extract the true room response from background noise.
In the z-domain, the poles and zeros of a real all-pass filter are reciprocal reflections of each other with respect to the unit circle. If a pole exists at , a zero exists at 2. All-Pass Phase Function
While it does not change the EQ balance, stacking these filters provides highly sought-after utility in modern sound design:
If you are looking for the "paper" behind the math or the project itself, here are the most relevant resources: allpassphase
Whether you are designing a reverb algorithm, correcting a loudspeaker’s time alignment, or simply trying to understand why your snare drum sounds "soft," the key lies in the phase. By learning to measure, design, and listen for allpassphase effects, you move from being a passive user of filters to an active sculptor of time itself.
The transfer function of a first-order digital all-pass filter takes the form: [ A(z) = \fraca + z^-11 + a z^-1, \quad |a| < 1 ]
This is the most common use case. Imagine you have a kick drum and a bass guitar playing the same note. Even if they are perfectly in time on the grid, the waveforms might be out of phase. This causes the low end to cancel out, making your mix sound thin and weak. When measuring a room’s impulse response, engineers use
Several key properties make all-pass filters indispensable:
If you are looking to design a specific type of filter or have questions about the mathematics behind them, please let me know:
For every benefit, there is a risk. Unchecked allpassphase implementation can ruin a mix instantly. Because the amplitude remains flat, beginners often add dozens of all-pass filters without realizing the damage. In the z-domain, the poles and zeros of
Constant at 1 (or any non-zero constant), meaning no frequency is made louder or quieter.
When an audio signal passes through an all-pass filter, different frequencies are shifted in time relative to one another. When multiple all-pass filters are stacked sequentially, this effect becomes heavily magnified. This phenomenon is known as or transient smearing .
