Table of Contents
Advertisement
In addition to these well-controlled and repeatable
signals, live speech may be used as a test signal. It
will, of course, be unequalized and at an uncontrolled
level.
3 Analysis of broad-band signals
Broad-band signals contain energy at many frequen-
cies simultaneously. Such signals are usually analyzed
in a series of narrow frequency bands to produce a
spectrum. The auditory system functions on a loga-
rithmic frequency scale and analyzes broad-band sig-
nals in critical bands which approximate 1/3 octave
bands (Figure 7). Using 1/3 octave bands for analysis
of broad-band signals allows measured levels to be
compared more readily to narrow-band behavioral
measures, such as threshold. For this reason, analy-
sis of broad-band signals in all On-ear measurements
(including Speechmap Test box mode) is in 1/3 octave
Figure 7: Analysis bandwidths of some real-ear analyzers and critical bands for a normal ear. If the analysis
bandwidth is less than a critical band, aided response curves shown to be at threshold (or UCL) will actually
be well above it.
bands. Other analyzers use narrower analysis bands,
sometimes having constant bandwidth. As shown in
Figure 7, this can result in significantly underestimat-
ing the audibility (or comfort or discomfort) of a com-
plex signal.
It should be noted that, in all On-ear measurements,
1/3 octave band SPL is displayed (and reported when
saving data to a file) at 1/12 octave frequencies which
provides curve smoothing and causes the spectrum
of narrow-band signals to appear broadened. When
calculating overall rms from these data, the SPL at 225
Hz and subsequent 1/3 octave increments should be
used.
Analysis of broad-band signals in all Test box tests,
except Speechmap, is in 1/12 octave bands and the
band SPL is displayed at 1/12 octave frequencies.
7
Advertisement
Table of Contents
