Acoustic test with reflected paths, 69 4.27. path length difference, Null for frequencies – Metric Halo SpectraFoo User Manual

Page 69

Advertising
background image

The Transfer Function

69

Under normal acoustic test conditions the response signal of the system will be the actual system response plus
noise. The noise (which can be electronic or acoustic), is, by definition, uncorrelated with the source signal.
The coherence of noise with any source signal (except an exact, time-aligned copy of the noise) is zero. As a
result, the noise that corrupts the response signal will tend to decrease the coherence of the measurement.

The fact that there is noise in the test environment is not normally a problem – when doing the test we arrange
for the test signal to be louder than the surrounding noise. Even if the signal is not louder than the environmental
noise at all times, the MBM thresholding allows the Transfer Function to only make measurements when the
source signal is louder than the environmental noise.

There is one situation that is problematic when making acoustic measurements. When the system has acoustic
reflections, the response signal may have hard or partial nulls at certain frequencies. These nulls depend on
the specific geometry of the acoustic environment (especially the location of the test microphone) and don’t
really represent the response of the system. A null is created at a specific frequency when multiple signal paths
cause the test signal to cancel out at the test mic position:

Reflective Acoustic Surface

(e.g. wall, table, ceiling, etc.)

Figure 4.26: Acoustic test with reflected paths

The reflected path introduces a copy of the signal with a time delay. The reflected signal takes longer to reach
the microphone than the direct signal because of the path length difference.

Path

Length

Difference

Direct Path Length

Reflection Point

Figure 4.27: Path length difference

If we call the path length ∆ l and the speed of sound V, the response at the microphone will have a null for
frequencies where n is any integer, as shown in this equation:

Figure 4.28: Null for frequencies

This is just a simple comb.

In any real acoustic environment there will be many reflective surfaces with many different reflectivities and
phase shifts, so we will not see a simple comb filter but a complex set of partial nulls. At the partial nulls in
the response at microphone position, little of the source signal will be detected; only the environmental noise
will be measured. This means that at the nulled frequencies, the coherence will be very low.

Advertising