Vox -01 Operating Manual page 46

Speech intelligibility
Hide thumbs Also See for VOX-01:
Table of Contents

Advertisement

6.3 Full STI
The Speech Transmission Index (STI) can be measured by means of various test methods.
The STIPA method, as implemented by the Quick STIPA and STIPA pro modules, is a fast
and efficient version of the STI originally developed for public address systems. Although
fast and robust, STIPA has its limitations: only 2 modulation frequencies are measured per
octave band, whereas the complete Speech Transmission Index measurement scheme calls
for 14 modulations per octave band. This means that STIPA is undersampled in the modu-
lation domain. For many systems, the measured STI is not (or hardly) affected. However,
some cases call for "Full STI" measurements:
Whenever echoes are present (delayed loudspeakers, large halls and cathedrals)
When tested systems or channels feature complex combinations of distortion factors.
Full STI measurements were available in commercial measuring tools up to the late 1990s,
but have subsequently been absent from the market for two decades. The fact that a full
STI measurement used to take around 15 minutes made Full STI measurements entirely
impractical. Also, the introduction of STIPA around 2002 made a reasonable alternative
available.
The Full STI module now available on the VOX-01 fully implements every aspect of
the Speech Transmission Index model, making it the most universally applicable speech
intel-ligibility metric to date:
Fully compliant with the Full STI definition according to IEC-60268-16
Computes the complete 14x7 Modulation Transfer Function matrix
Uses an efficient algorithm that completes the measurement in approx. 1 minute
Implements the direct measurement method as defined in IEC-60268-16
The "direct method" (in which a modulated test signal is presented to the channel under
test), as implemented here, has crucial advantages over the "indirect method" (where the
STI is estimated based on a measured impulse response). The indirect method implicitly
assumes that the channel under test is a linear time-invariant system. Whenever electro-
acoustic components are involved, the tested channel never fully satisfies the linearity re-
quirement. Indirect estimates are notoriously inaccurate in many real-life scenarios, but
have nonetheless been used extensively, for lack of alternatives based on the direct method
(such as this module).
The Full STI module implements a novel measurement algorithm, that leverages the
ad-vanced signal processing capabilities of the VOX-01. This algorithm, which is
documented separately in an Embedded Acoustics white paper, has: he following
characteristics:
46
New in v2.0
rt
VOX-01 Operating Manual

Advertisement

Table of Contents
loading

Table of Contents