FFT, Windowing, and Impulse Response
The math behind the spectrum display. Deep dive into how data is transformed.
Core Concepts: FFT vs RTA
- Select Measurement Settings: Go to Options > Measurement Settings.
- Choose Analysis Type: Select "FFT" for a spectrum or "RTA" for Real Time Analyzer.
- Set FFT Length: In the FFT settings, change the length. Default is often 2048 or 4096 samples.
The Mathematics of Resolution
Frequency Resolution Formula
The precision of your frequency bins is determined by this equation:
Resolution = Sample Rate / FFT Length
- Shorter FFT: Better time resolution (you see the decay of a sound faster), but worse frequency resolution. The graph looks "blurry" at low frequencies.
- Longer FFT: Better frequency resolution (sharper peaks visible). However, you lose time-domain information. You can't easily isolate the first reflection from the decay if the FFT is too long.
Pro Tip: To see a distinct peak at 20Hz using an FFT of 48kHz, you need an FFT length that results in bins smaller than 20Hz. A standard 1024-sample FFT at 48kHz yields ~46Hz resolution. You must use a longer FFT (e.g., 512 samples) to see fine detail below 100Hz.
RTA vs Sweep
| Method | What it gives |
| FFT | Transfer function + Impulse response (Snapshot) |
| RTA | Live spectrum / fractional-octave response (Real-time) |
Windowing
Selecting the Window
Windowing** means selecting only part of the impulse response before FFT calculation.
- The IR:** The Impulse Response contains Direct Sound (t=0) and Reflections (later). A window cuts this off.
- Left Window: Before the IR peak. Isolates direct sound but limits low-frequency resolution.
- Right Window: After the IR peak. Includes room effects.
Frequency Resolution Rule: Frequency resolution is roughly 1 / window duration. A 10 ms window cannot show valid data below about 100 Hz. If you need to measure a bass peak at 50Hz, your IR window must be longer than ~20ms.