Question 1

What is a transfer function measurement in audio?

Accepted Answer

A transfer function measures how a system (amplifier, speaker, room, or the entire signal chain) modifies an input signal. It is computed as H(f) = Y(f) / X(f), where Y(f) is the output spectrum and X(f) is the input spectrum. The result shows the magnitude response (gain or loss at each frequency) and phase response (time delay at each frequency). Unlike a simple spectrum analysis (RTA), which shows what is happening, a transfer function shows what the system is doing to the signal. This distinction is critical for sound system tuning because the transfer function isolates the system behavior from the program material. SonaVyx uses the H1 estimator for transfer function measurement, which minimizes noise at the output.

Question 2

What is the H1 estimator and why does SonaVyx use it?

Accepted Answer

The H1 estimator computes the transfer function as H1(f) = Gxy(f) / Gxx(f), where Gxy is the cross-spectral density between input and output, and Gxx is the auto-spectral density of the input. H1 minimizes the effect of noise at the output (microphone position), which is the dominant noise source in acoustic measurements. The alternative H2 estimator (H2 = Gyy/Gyx) minimizes noise at the input and is better when the excitation signal is contaminated. For live sound measurement where the source signal is clean but the microphone picks up ambient noise, H1 is the correct choice. Welch averaging over multiple FFT blocks improves the H1 estimate by reducing variance. SonaVyx applies Welch averaging with 50% overlap by default.

Question 3

What does coherence tell me about my measurement quality?

Accepted Answer

Coherence gamma-squared ranges from 0 to 1 and indicates how much of the output signal is linearly related to the input signal at each frequency. Coherence of 1.0 means the output is perfectly determined by the input (no noise, no nonlinearity). Coherence of 0.5 means only half the output energy at that frequency comes from the input. Low coherence can be caused by: ambient noise, reflections from nearby surfaces, time-varying conditions, nonlinear distortion, or the system not being excited at that frequency. As a rule of thumb: above 0.85 is reliable, 0.5-0.85 is usable with caution, below 0.5 should not be trusted. SonaVyx displays coherence as a colored bar chart (green/amber/red) below the magnitude plot.

Question 4

What is the difference between a transfer function and an RTA measurement?

Accepted Answer

An RTA (Real-Time Analyzer) measures the absolute spectrum of whatever signal the microphone captures. It shows what the audience hears, including the program material spectrum. A transfer function measures the ratio of output to input, revealing how the system modifies the signal regardless of what is being played. This means you can use any broadband signal (pink noise, music, speech) as the excitation and still see the system response. An RTA of a system playing a guitar solo will show the guitar spectrum, not the system response. A transfer function during the same guitar solo shows how the system is coloring the guitar. For EQ decisions, you need the transfer function. For monitoring overall levels and checking for obvious problems, RTA is sufficient.

Question 5

Do I need a reference signal for transfer function measurement?

Accepted Answer

Yes. A transfer function requires both an input (reference) signal and an output (measurement) signal. There are several ways to obtain the reference: direct electrical feed from the mixing console or DSP output via a DI box or audio interface, which gives the cleanest reference. Loopback mode where SonaVyx generates the test signal and uses it as the reference internally. Two-microphone method where one mic is placed near the source as the reference and the other at the measurement position. SonaVyx supports loopback mode for quick single-device measurements and multi-device mode where one device plays the signal and another captures it. For live event tuning, a direct console feed is preferred because it lets you measure during the show using program material.

Question 6

What FFT size should I use for transfer function measurement?

Accepted Answer

FFT size determines the frequency resolution and time resolution trade-off. Larger FFT gives finer frequency resolution but slower update rate. At 48 kHz sample rate: 4096 points gives 11.7 Hz resolution (good for broadband overview), 8192 points gives 5.9 Hz resolution (good general purpose), 16384 points gives 2.9 Hz resolution (good for identifying room modes below 200 Hz), and 32768 points gives 1.5 Hz resolution (for detailed low-frequency work). For live sound system tuning, 8192 is the standard starting point. Increase to 16384 when investigating low-frequency problems. Apply smoothing (1/6 or 1/3 octave) to reduce visual noise while maintaining the underlying resolution. SonaVyx defaults to 8192 with 1/6 octave smoothing.

Question 7

What is phase in a transfer function and why does it matter?

Accepted Answer

Phase represents the time delay the system introduces at each frequency. A linear phase response (constant group delay) means all frequencies are delayed equally, preserving the waveform shape. Real systems have non-linear phase due to: crossover filters in speakers (which cause rapid phase rotation), room reflections (which create interference patterns), and DSP processing latency. Phase matters because when multiple speakers reproduce overlapping frequency ranges, their phase relationship determines whether they add constructively (in phase) or destructively (out of phase). At crossover frequencies where two drivers overlap, phase alignment within ±30 degrees ensures smooth summation. SonaVyx displays unwrapped phase and group delay alongside the magnitude response.

Question 8

Can I measure a transfer function during a live show?

Accepted Answer

Yes, this is one of the most powerful uses of transfer function measurement. Feed the console output as the reference signal to SonaVyx (via a split or direct out) and capture the room response with a measurement microphone. The transfer function will show the system response regardless of what music is playing because the H1 estimator divides out the program content. Requirements: the program material must have broad spectral content (most music works, speech is marginal). Averaging is essential to build up coherence across all frequencies. Use at least 4-8 averages. Very narrow-band music (solo instruments, electronic bass) may not excite all frequencies, resulting in low coherence in unexcited bands. SonaVyx shows coherence so you know which frequency ranges have valid data.

Transfer Function Measurement Basics: H1, Coherence, Phase, and Magnitude Explained

Frequently Asked Questions