A system-response approach to FIR correction

FIRForge is a precision measurement and FIR correction system for live and studio audio work. It is designed to measure how a loudspeaker system behaves in its actual acoustic environment, then create a correction filter that brings the measured response closer to a target you choose.

The goal is simple: hear more of the speaker system itself, and less of the tonal problems introduced by placement, boundaries, system tuning, and the surrounding environment.

FIRForge is built for situations where speed and trust matter. A touring engineer may be working with a different PA system and venue every night. A studio engineer may be working in a room that is treated, but not perfect. In both cases, the problem is the same: the loudspeaker system does not reach your ears as a neutral, predictable reference. The measured response includes the speaker, its placement, the listening or coverage area, and the acoustic environment around it.

FIRForge approaches this as a system-response problem.

Instead of relying on a single microphone position, FIRForge uses continuous sine sweep measurement while the microphone is moved through the listening or coverage zone. This captures many positions and creates a spatially averaged response. The purpose is not to chase one perfect point in space, but to build a practical representation of how the system behaves across the area that matters.

A single measurement point can be misleading. Move the microphone slightly and the measured response may change dramatically, especially in the low and low-mid range where boundaries, standing waves, cancellations, and placement effects dominate. By averaging many positions, FIRForge reduces the influence of hyper-local anomalies and derives a correction target that is usually more stable, more forgiving, and more musically useful.

FIRForge displays this averaged result as the system's Power Response. In FIRForge, Power Response means the spatially averaged response gathered from the measured zone: a practical view of the system's tonal behavior over that area. It includes the effect of the direct sound, placement, boundaries, reflections, and the surrounding acoustic environment.

It is not meant to replace every possible acoustic analysis view. It is the response view FIRForge uses because it is the most useful foundation for fast tonal correction.

Minimum-phase describes a system where the magnitude response and phase response are connected. When the magnitude response changes, the phase changes with it. A peak, dip, resonance, or roll-off is not only a change in level. It also has a time/phase behavior attached to it.

Many real loudspeaker and EQ behaviors are minimum-phase or have a strong minimum-phase component. This is why traditional analog EQs and most IIR filters work the way they do: when you boost or cut, you also get the corresponding phase movement. That phase movement is not a bug. It is part of the correction.

A minimum-phase filter is also the most time-compact version of a filter for a given magnitude response. In simple terms, its energy is pushed as early as possible in the impulse response. It does not need to be centered in time like a linear-phase FIR filter.

This matters because linear-phase FIR correction has a cost. A linear-phase FIR filter can keep phase perfectly linear, but it does this by using a symmetrical impulse response. That symmetry requires latency, often half the filter length, and can create pre-ringing around transients. This can be useful in some offline or mastering situations, but it is usually the wrong tradeoff for live sound and real-time monitoring.

FIRForge uses minimum-phase FIR correction because it gives precise FIR magnitude control while keeping the filter practical for real-time playback. The correction can reshape broad tonal problems and apply the phase behavior that naturally belongs with that magnitude correction, without adding the delay or pre-ringing behavior associated with linear-phase FIR.

In other words, FIRForge is not trying to separate magnitude from phase. It is building a correction filter where magnitude and phase move together in the natural minimum-phase relationship. The result is precise tonal correction with a compact impulse response, natural transient behavior, and real-time playback through FIRForge Convolver with 0 samples of added processing delay.

Measurement position still matters. Moving-mic averaging does not mean moving the microphone randomly through the room. It means deliberately scanning the part of the sound field that should represent the listening or coverage area.

Loudspeakers are not omnidirectional point sources. Their radiation pattern changes with frequency. At low frequencies, energy spreads widely and the acoustic environment has a strong influence. At higher frequencies, speakers become more directional, and off-axis response can roll off or become irregular.

This means the measurement area must be chosen carefully. If the microphone is moved too far outside the useful listening axis or coverage zone, the measurement may capture less direct high-frequency energy and more reflected or off-axis energy. A filter based on that information may look reasonable on a graph but sound wrong in use.

Spatial averaging works best when every measured position still belongs to the area you actually want the correction to serve. The goal is not to average the entire room or venue. The goal is to capture the useful sound field of the system.

For compact speakers and studio monitors, a useful starting point is often a distance of roughly two to three times the speaker's largest physical dimension. This usually places the microphone far enough away for the drivers to integrate, while still keeping the measurement focused on the useful listening area. For PA systems, fills, subs, and larger deployments, the measurement area should be chosen according to the real coverage zone being corrected.

During measurement, the microphone should be moved deliberately through the area that matters. Think of it as scanning the useful sound field: slightly up and down, left and right, forward and backward, without wandering into positions that do not represent the intended listening or coverage zone.

FIRForge can also skip sweeps from the beginning and end of a measurement pass, so the useful part of the capture can be isolated after the engineer has walked to and from the measurement position.

Once the averaged system response is captured, FIRForge creates a correction filter. The engineer defines the correction range, chooses the FIR tap length, sets the output sample rate, and shapes the target if needed. Shorter filters can be used for lighter or higher-frequency correction. Longer filters provide finer low-frequency correction resolution when the system requires it.

The correction filter is then deployed through FIRForge Convolver, the playback engine of the system. The convolver is designed for live and studio use, adding 0 samples of processing delay even with long FIR filters. It also provides practical deployment tools such as target curve shaping, automatic input gain compensation, automatic L/R balance, input-compensated bypass, preset recall, polarity control, delay, and crossover-style filter tools.

This is the core FIRForge idea: measure the real system, average the area that matters, correct the parts that are safe and useful to correct, deploy the result with no added processing delay, and keep the workflow fast enough for real work.

FIRForge is built around a focused philosophy: measure the response that matters, create a correction that can be trusted, and deploy it quickly.

The system can become as technical as the work requires, but the core workflow stays simple: measure, average, correct, deploy, and listen. FIRForge is designed to keep the engineer close to the sound instead of buried in unnecessary steps.