IEC 60268-16 (STIPA)

IEC 60268-16 defines the speech transmission index, STI, and the methods for measuring it, including STIPA, the method built into many sound level meters and analysers. It belongs to the sound-system equipment series rather than to the sound level meter standards, but it has become an instrument specification in its own right because voice-alarm and public-address commissioning require a measured STI, and the meter is the tool that delivers it.

The speech transmission index

STI is a single number between 0 and 1 that expresses how well a transmission path, from talker or loudspeaker to listener, preserves the intelligibility of speech. It is derived from the modulation transfer function: speech is modelled as a carrier in each of seven octave bands from 125 Hz to 8 kHz, modulated at fourteen frequencies between 0.63 and 12.5 Hz. Reverberation, background noise, echoes and distortion all reduce the depth of the received modulation, and the loss in each band and modulation frequency is converted to an apparent signal-to-noise ratio, weighted and summed. Full STI uses all 98 combinations and takes several minutes per point.

The standard includes qualification bands for interpreting the result. For public address systems, values are banded from A+ (above 0.76) through A to J down to U (below 0.36), with 0.50 widely used as the minimum for emergency announcements and 0.45 to 0.50 the common requirement in fire codes. These are categories in the standard, not legal limits; the applicable code sets the requirement.

STIPA

STIPA, the speech transmission index for public address systems, is the practical measurement method. A test signal with two modulation frequencies per octave band, applied simultaneously, is played through the system under test from a file or a signal generator, and the analyser measures the received modulation depth in each band over about 15 to 20 seconds. A single point therefore takes well under a minute. STIPA assumes a linear, time-invariant system; it is not valid for paths with strong non-linear distortion or for vocoded and some digital radio links, for which the standard gives guidance.

Because the result depends on both the system and the room, STIPA is measured at listener positions with the system at its operating level and with representative background noise. Where noise is not representative at the time of the test, the standard allows the measured STI to be corrected using a separately measured noise spectrum, and instruments typically offer that correction. The signal source matters: the test signal must be the one specified in the edition being used, at the correct spectrum and level.

Editions

Results from different editions are not identical for the same system, chiefly because of the masking corrections, and a specification or code may require a particular edition. Instruments in the database cite revision 4, revision 5 or both; the statement is reproduced as the manufacturer gives it. Firmware updates sometimes move an instrument from one edition to the next, so the firmware version matters for older instruments.

What the instrument needs

STIPA is a software function on a sound level meter or analyser with octave-band filters. The datasheet should state the edition implemented, whether the function is standard or an option, whether a test-signal file or generator is supplied, and whether noise correction is available. The microphone should be class 1 to IEC 61672-1 and the filters class 1 to IEC 61260-1, as the standard recommends, although STIPA itself is not a classed measurement. Averaging several measurements per position is recommended because the test signal is stochastic.

Instruments offering STIPA

The table lists every instrument in the database whose manufacturer cites IEC 60268-16 or states a speech intelligibility function, with the statement as given. For most it is an optional licence; the instrument page says what the manufacturer publishes about the option. The glossary defines STI and STIPA.