What FFT size should I use for audio measurement?

A 4096-point FFT at 48 kHz provides 11.7 Hz resolution and approximately 85 ms time blocks, offering a good balance between frequency detail and update speed. Use 8192 or 16384 points for low-frequency analysis below 100 Hz where finer resolution reveals room modes and narrow resonances.

Why does RTA use pink noise instead of white noise?

Pink noise has equal energy per octave band, matching the logarithmic frequency spacing of RTA displays and human hearing. White noise has equal energy per Hz, which makes higher octave bands appear louder on an RTA display. Pink noise produces a flat RTA trace through a system with flat frequency response.

What is spectral leakage and how do window functions fix it?

Spectral leakage occurs because FFT assumes the signal is periodic within the analysis block. Discontinuities at block boundaries spread energy across bins. Window functions taper the signal amplitude toward zero at the edges, reducing these discontinuities. The tradeoff is slightly reduced frequency resolution.

Is 1/3 octave resolution sufficient for system EQ?

One-third octave resolution suffices for broadband system equalization using graphic EQ. For parametric EQ targeting narrow resonances, feedback frequencies, or room modes, narrowband FFT analysis is necessary. Most sound engineers use 1/3 octave RTA for overall tonal balance and FFT for surgical corrections.

What FFT size should I use for audio measurement?

A 4096-point FFT at 48 kHz provides 11.7 Hz resolution and approximately 85 ms time blocks, offering a good balance between frequency detail and update speed. Use 8192 or 16384 points for low-frequency analysis below 100 Hz where finer resolution reveals room modes and narrow resonances.

Why does RTA use pink noise instead of white noise?

Pink noise has equal energy per octave band, matching the logarithmic frequency spacing of RTA displays and human hearing. White noise has equal energy per Hz, which makes higher octave bands appear louder on an RTA display. Pink noise produces a flat RTA trace through a system with flat frequency response.

What is spectral leakage and how do window functions fix it?

Spectral leakage occurs because FFT assumes the signal is periodic within the analysis block. Discontinuities at block boundaries spread energy across bins. Window functions taper the signal amplitude toward zero at the edges, reducing these discontinuities. The tradeoff is slightly reduced frequency resolution.

Is 1/3 octave resolution sufficient for system EQ?

One-third octave resolution suffices for broadband system equalization using graphic EQ. For parametric EQ targeting narrow resonances, feedback frequencies, or room modes, narrowband FFT analysis is necessary. Most sound engineers use 1/3 octave RTA for overall tonal balance and FFT for surgical corrections.

FFT vs RTA — When to Use Each for Audio Analysis

FFT Analysis Fundamentals

The Fast Fourier Transform converts time-domain audio samples into frequency-domain magnitude and phase data. An FFT of size N at sample rate fs produces N/2+1 unique frequency bins spaced at fs/N Hz apart. A 4096-point FFT at 48 kHz yields 2049 bins with 11.7 Hz resolution. Larger FFT sizes provide finer resolution but slower update rates.

Windowing and Spectral Leakage

Applying a window function before the FFT reduces spectral leakage caused by analyzing finite-length signal blocks. Common windows include Hann (good general purpose), Blackman-Harris (excellent sidelobe rejection at -92 dB), and Flat-Top (best amplitude accuracy). The choice affects frequency resolution and amplitude accuracy inversely.

RTA and Octave Band Analysis

RTA per IEC 61260-1 groups spectral energy into standardized bands. Full octave bands follow the ratio 1:2. One-third octave bands provide three bands per octave. IEC 61260-1 clause 5.2 specifies center frequencies using the base-2 formula. Class 1 filters must maintain attenuation within tight tolerances at the center and band edge frequencies.

When to Use FFT

Use narrowband FFT analysis for identifying specific tonal components, feedback frequencies, hum harmonics, and equipment noise. Transfer function measurements require FFT resolution to see phase response and coherence at individual frequencies. FFT is essential for identifying comb filtering patterns where the null spacing reveals the time delay.

When to Use RTA

RTA excels at broadband spectral monitoring during live events. Octave and third-octave displays match how we perceive sound energy distribution. Room EQ decisions using pink noise are best made with 1/3 octave RTA because the constant percentage bandwidth matches the logarithmic frequency response of the ear.

Combining Both Approaches

Professional measurement workflows use both. Start with RTA to identify broad spectral trends, then switch to narrowband FFT to diagnose specific problems. Modern tools like SonaVyx display FFT with an octave-band overlay, letting you see both simultaneously. The key is matching the analysis resolution to the question you need answered.