What is the difference between H1 and H2 estimators?

H1 = Gxy/Gxx minimizes output noise (mic ambient noise), ideal for acoustic measurement. H2 = Gyy/Gyx minimizes input noise. H1 is the standard for sound system measurement per AES-2id:2023 clause 4.1 because the microphone signal is noise-contaminated.

What coherence value indicates reliable data?

Above 0.85 is highly reliable. Between 0.5-0.85 is moderately useful. Below 0.5 is dominated by noise or non-linearity. Never apply EQ based on magnitude data at frequencies where coherence is low because the data is unreliable at those frequencies.

Why does phase matter for alignment?

Phase differences at crossover determine speaker summation. 180-degree offset causes cancellation. Transfer function phase reveals alignment issues. Adjusting delay and polarity to achieve phase coherence at crossover ensures constructive power summation between drivers.

How many averages are needed?

AES-2id recommends minimum 8 spectral averages for stable coherence. More improves reliability. For noisy environments, use 16-32. The tradeoff is time: each average needs one FFT block of signal. Balance speed with measurement quality.

What is the difference between H1 and H2 estimators?

H1 = Gxy/Gxx minimizes output noise (mic ambient noise), ideal for acoustic measurement. H2 = Gyy/Gyx minimizes input noise. H1 is the standard for sound system measurement per AES-2id:2023 clause 4.1 because the microphone signal is noise-contaminated.

What coherence value indicates reliable data?

Above 0.85 is highly reliable. Between 0.5-0.85 is moderately useful. Below 0.5 is dominated by noise or non-linearity. Never apply EQ based on magnitude data at frequencies where coherence is low because the data is unreliable at those frequencies.

Why does phase matter for alignment?

Phase differences at crossover determine speaker summation. 180-degree offset causes cancellation. Transfer function phase reveals alignment issues. Adjusting delay and polarity to achieve phase coherence at crossover ensures constructive power summation between drivers.

How many averages are needed?

AES-2id recommends minimum 8 spectral averages for stable coherence. More improves reliability. For noisy environments, use 16-32. The tradeoff is time: each average needs one FFT block of signal. Balance speed with measurement quality.

Transfer Function in Audio — Magnitude, Phase & Coherence

Transfer Function Fundamentals

H(f) reveals magnitude (dB), phase (degrees), and coherence. AES-2id clause 4.1 defines H1(f) = Gxy(f)/Gxx(f) using cross-spectral and auto-spectral densities. H1 minimizes output noise effects, ideal for acoustic measurement where ambient noise contaminates the microphone.

Magnitude Response

Flat magnitude (±3 dB, 80 Hz to 16 kHz) indicates transparent system. Deviations show EQ, speaker roll-off, room resonances. Smoothing (1/3 to 1/24 octave) reveals trends. Always check unsmoothed data first for narrowband issues.

Phase and Group Delay

Constant unwrapped phase slope = pure delay. Deviations reveal filter delay, crossover effects, reflections. Group delay = negative derivative of phase. Flat group delay means simultaneous frequency arrival. Crossover filters create group delay peaks.

Coherence

Gamma-squared ranges 0-1 per AES-2id clause 4.3. Above 0.85 = reliable data. Low coherence = noise, distortion, or time-variance. Never EQ where coherence is low. Increase with more averages, higher signal level, or quieter conditions.

Measurement Setup

Reference signal through system, capture input and mic output simultaneously. Welch's method with 50-75% overlap. Minimum 8 averages per AES-2id clause 5 for stable coherence estimation.

Transfer Function — Magnitude, Phase & Coherence

Transfer Function Fundamentals

Magnitude Response

Phase and Group Delay

Coherence

Measurement Setup

Frequently Asked Questions

Related Articles

Related Articles

→FFT vs RTA — When to Use Each for Audio Analysis

→Comb Filtering — Causes, Detection & Solutions in Audio

→How to Read & Measure Frequency Response

→Speaker Time Alignment — Phase Coherence & Summation