Core Measurement Theory and Transfer Functions
Understand what REW actually measures before you trust a curve.
What this is for
REW primarily uses a logarithmic sine sweep to measure the transfer function of an audio system. The sweep is played through the output, captured at the input, and processed via FFT to separate the linear response from distortion and noise. This gives far better signal-to-noise ratio than a click and enables time-domain analysis.
Before touching controls, you must understand the difference between a transfer function and a live RTA spectrum. A transfer function is valid only for one source position, one microphone position, and one set of settings.
Key concepts
- Signal, sample rate, bit depth, full scale: A signal is a sequence of samples. The highest capturable frequency is half the sample rate. Bit depth sets resolution. Full Scale (FS) is the maximum digital level; exceeding it clips.
- Spectrum vs transfer function: A spectrum shows the frequencies in a signal. A transfer function shows how a system changes amplitude and phase at each frequency.
- FFT and RTA: The FFT converts time data to frequency data. An RTA groups FFT bins into fractional-octave bands, closer to human hearing, but cannot separate direct sound from reflections.
- Impulse response (IR): The time-domain description of a system. It contains direct sound, reflections, decay, and distortion components at negative time when measured by a log sweep.
- Windowing: Selecting only part of the IR before FFT. Short windows isolate direct sound but limit low-frequency resolution. Long windows include room effects.
- Minimum phase, excess phase, group delay: A minimum phase system has the lowest possible delay for a given magnitude response and can be inverted. Excess phase is the difference between measured phase and minimum phase. Flat excess group delay indicates a minimum phase region.
- Measurement validity: A measurement is valid only for the exact source/mic position, level, sample rate, calibration, IR window, and processing state.
Concrete steps to apply in REW
Outputs and interpretation
| Graph | What it tells you |
|---|---|
| SPL & Phase | Frequency response magnitude and phase. Peaks/dips show modal or boundary effects. |
| Impulse | Time-domain response: direct sound, reflections, decay, distortion at negative time. |
| Waterfall / Spectrogram | How frequency content changes over time; reveals modal ringing. |
| RTA | Live combined spectrum of source + room. Cannot separate direct sound from reflections. |
Worked example: Before EQ, check if a low-frequency peak is minimum phase
Measure the listening position with a full-range sweep. Open Group Delay, generate a minimum phase version, and view excess phase / excess group delay. If the excess group delay is flat in the region of a peak, the peak is likely minimum phase and can be cut with EQ. If it is not flat, EQ will not fully correct the problem.
Common mistakes
- Thinking an RTA measurement is a full transfer function.
- Assuming one measurement is valid everywhere in the room.
- Ignoring phase and time-domain behavior.
- Believing EQ can fix every acoustic problem.