Background
These audio files scan all audible frequencies from 20 Hz to 20 kHz
in a 20 s time span. When the time scale is logarithmic, sweeping
between 20 Hz and 40 Hz (one octave) or from 10 kHz to 20 kHz (one
octave) will take the same time. When the time scale is linear,
sweeping from 20 Hz to 40 Hz will be much faster, and equal to the
time taken to sweep from 10 kHz to 10.02 kHz (which only represents
a 3-cent pitch interval!).
A log sweep is characterized by a pink-like spectrum, a
linear sweep by a white-like
spectrum.
Applications
Sine sweeps are used as reference tones to check frequency
response or the adverse effects of room modes. Compared to
pink noise
or
white noise,
sine sweeps produce frequencies with a much
higher energy, as they dedicate the fully available dynamic to
play one frequency at a time, not simultaneous frequencies like noise.
This gives sine sweeps a better immunity against room ambience and
background noises. Sine sweeps are particularly useful to determine
resonant frequencies.
The sound files
|
↓
|
↓
|
Logarithmic (20Hz-20kHz, -3dBFS, 20s) |
Linear (20Hz-20kHz, -3dBFS, 20s) |
When checking by ear, the sound should evolve smoothly from the
lowest frequency to the highest. No strong frequency dips or peaks
should be present.
Please note that due to the increased
sensitivity of our ears in some area of the audio spectrum, some
frequency ranges - particularly in the upper medium - may be
perceived louder than they actually are (the so-called
Fletcher-Munson effect). For this test, don't pay attention to
this phenomenon, but rather to strong and localized frequency dips
or peaks. When checking by ear, use the logarithmic version, as
our ears are more or less working logarithmically.
In combination with an audio spectrum analyzer, "peak hold" is
necessary to measure the full frequency scale in a single session.
Assuming the spectrum analyzer performs its analysis across the
log-frequency scale, perfect systems will exhibit a flat response
to the logarithmic swept sine stimulus. When using a linear FFT
analyzer, please refer to the linear version instead.