SPEECH INTELLIGIBILITY / PRIVACY
MEASUREMENT SYSTEMS

Equipment You Need To Accurately Measure:

  • Updated information regarding Mass Notification Systems is available. Click here to learn more.

  • STICistm system used at New England Patriot's Gillette Stadium. Click here to read about it.

  • STICistm system used at Boston Convention & Exhibition Center. Click here to read about it.

  • Meets or Exceeds International Standards ref: ISO 9921, E366.803, IEC 60268-16, 60849, ANSI S3.05.

  • Test CD Developed by TNO Human Factors in Holland - Developer of STI in Europe.

  • DSP Series Analyzers and companion TALKBOX Signal Generator Developed by Gold Line.

STI Measurement Technology Covered By
U.S. Patent 6,792,404

INTELLIGIBILITY / PRIVACY SYSTEM COMPONENTS
DSP2B

DSP2B/DSP2BP - DSP2B Dedicated Speech Intelligibility Meter Measures Sound Pressure Level and Speech Intellegibility Index in CIS or STI formats using the OPTion STICistm Software. DSP2BP includes OPTion PI Privacy Index.

DSP30B

DSP30B/DSP30BP / DSPCIB/BP - Full Function Stand Alone Audio Spectrum Analyzers Measure Sound Level and Speech Intellegibility Index in CIS or STI formats when used with OPTion STICistm Software. DSP30BP includes OPTion PI Privacy Index.(DSP30B Shown.)

TALKBOX

TALKBOX - Signal Generator - An Integral Part of The Speech Analysis System - Used with either DSP2B/BP, DSP30B/BP or DSPCIB/BP - Applies Specialized Test Tones To A Voice Message System or selected enviroment.spacer52.gif - 248 Bytes


Go To YouTube DirectoryWatch a video about Intelligibility Topics. Read about Intelligibility Read about Privacy User Comments

Major International, European, and North American fire alarm codes now require that voice alarm systems meet minimum levels of speech intelligibility. The new codes and standards have spawned the need for education about the basics of intelligibility, and for tools to ensure that voice alarm systems are designed for good intelligibility, can be measured for compliance at the time of commissioning, and can be periodically tested during the life of the system.

The basics of speech intelligibility, including the factors that affect intelligibility in typical situations are reviewed in a white paper presented at the National Fire Protection Association Congress by Bose® Professional Systems. In addition, the various methods for measuring speech intelligibility are discussed, together with techniques for predicting intelligibility in advance of building construction or equipment installation. Finally, many of the practical considerations fire alarm professionals will face in managing the intelligibility requirement are presented.

To obtain a copy of this white paper, you can download it here.


Improving the Intelligibility of Safety and Security Announcements Using the DSP with OPTion STICistm

WHAT IS INTELLIGIBILITY? Speech intelligibility is not a physical quantity like Amperes, Volts, or BTU's. It is the degree to which we understand spoken language. Speech is not necessarily intelligible simply because it is audible. Everyone has experienced speech signals that are loud enough but overly reverberant, suffer from echoes or distortion and therefore are not understandable. Announcements in airports, train stations and places of worship often suffer from this problem.
MEASURING INTELLIGIBILITY WITH STI The Speech Transmission Index (STI) is a quantitative method of measurement covered in the International Electrotechnical Commission (IEC) standard 60268-16.
SPEECH TRANSMISSION INDEX (STI) METHOD The military and others began funding the research and development of a quantitative method of measuring speech intelligibility that could measure a system that had virtually any combination of the factors commonly found to affect speech intelligibility. The work that eventually led to what is now called the Speech Transmission Index method was started in the 1970's at a well-known research laboratory called TNO Human Factors in the Netherlands, mainly under NATO funding. The method developed proved robust under a wide array of conditions from a talker in a room to a telephone system to an area PA system. The STI method has been implemented in a number of commercially available devices and is today an international standard used extensively throughout the world. The STI method has an advantage in that it correctly accounts for all of the factors in the talker-to-listener transmission path that affect intelligibility. The method is based on replacing speech with a repeatable signal that has the same characteristics as actual speech.
MODEL DSP30 WITH THE STICistm INTELLIGIBILITY OPTION Until now, the routine measurements of intelligibility necessary to enforce the codes and standards were difficult to make. With this problem in mind, we set out to develop a dedicated instrument that would accurately measure speech intelligibility according to international standards, and do so quickly and simply, thus making it possible for experts and non-experts alike to quickly measure intelligibility. Their results show clearly the feasibility of such an instrument.

The DSP30 is a DSP-based 1/3 octave audio spectrum analyzer and sound pressure level meter. The analyzer is handheld, can be battery operated and is extremely easy to use. In addition, the DSP30 can be enhanced with a series of software/hardware options. For measuring intelligibilty, Option STICistm is available.

This option includes the STI-PA signal developed by TNO Human Factors in the Netherlands where the STI method was researched and developed. The STI-PA signal, contained on an audio CD, is input to the system under test.

STI measurements can be made by most without significant training. Once the STI-PA signal has been input to the system under test, a single button push on the DSP30 will produce the STI, converted to the Common Intelligibility Scale as specified in IEC standard 60849, in about 15 seconds. The STI-PA signal can be left continuously playing, making it possible to simply move the DSP30 to a new position and a new measurement initiated.

The DSP30 with STICistm software was tested to ensure that it behaved according to theory and met the goals for accuracy, portability, and simplicity. STI values obtained on the meter were compared to STI values obtained using the full 98 combinations of modulated noise as calculated on a reference system at TNO. A number of representative test conditions were used, including different bandpass conditions, noise levels, non-linear effects, reverberation and echo profiles. The results are shown in a white paper presented at the Reproduced Sound 17 Conference . The data show conclusively that the meter is accurate in a wide range of conditions typical of public address systems.

You can download a copy of this white paper here.

Gold Line equipment and software has been approved by TNO for the objective prediction of speech intelligibility.

You can download a copy of this letter here.

The Measurement of Speech Intelligibility
The draft version of the new ISO 9921 standard on the "Assessment of Speech Communication" defines speech intelligibility as: "a measure of effectiveness of understanding speech". This contribution describes and compares several of these measures for determining the intelligibility of a given speech transmission system. It may include the acoustical environment at the speaker and the listener position.
In general two principally different assessment methods may be applied:
(1) Subjective assessment, based on the use of speakers and listeners,
(2) Objective assessment based on physical parameters of the transmission channel.
A paper discussing "The Measurement of Speech Intelligibility" can be downloaded here.


Measuring and Improving Privacy Using the DSP with OPTion PI

WHAT IS PRIVACY? The Privacy Index is essentially the inverse of the intelligibility Index. One can think of it as an Un-Intelligibility Index. Whereas Gold Line's intelligibility firmware tells you when people are able to understand speech from a sound system, OPTion PI does the opposite. It tells you when speech is sufficiently blocked or masked so that people cannot understand the content. By doing so it predicts when various degrees of privacy have been achieved.
WORKING WITH INTELLIGIBILITY AND PRIVACY INDEX ISSUES Increasingly Codes and liability law are compelling modern consultants, contractors and AHJ's to find a way to verify that Voice Evacuation and Sound Masking systems actually perform as intended. The National Fire Protection Association (NFPA) has taken the important step of requiring intelligibility in every voice alarm system. Thus a code-based mechanism now exists ensuring that when a voice alarm system is used, we can be certain that its effectiveness will not be compromised because of poor speech intelligibility. See Appendix to NFPA 72. And, the new HIPAA standards for Patient Information privacy is driving the trend towards PI as the best method for assuring that masking systems are in fact providing adequate levels of privacy.
DSP ANALYZERS WITH THE OPT PI PRIVACY OPTION The firmware to upgrade a Gold Line digital real time analyzer to do PI analysis is called OPT PI. In order to accept the OPT PI firmware, an analyzer must also have our intelligibility software, OPT STICis. So the two packages, working together, allow a user to make Privacy Index Measurements. Gold Line analyzers that can accept OPT PI include the DSP30, DSPCIW and the DSP2.
MEASURING PRIVACY WITH PI White papers from Government Studies have recommended PI as the preferred method for assuring that speech remains confidential. Recent studies by Susan Miller, Esq. and David Sykes, PhD, have documented the superiority of Privacy Index measurements, and the legal issues related to failure to provide speech privacy under the new HIPAA medical regulations.
White Papers by Miller and Sykes:
Oral Communications - Myths and Facts.
Oral Communications - Myths and Facts.(Powerpoint)
A COMPARISON OF STI AND PI CONVERSIONS Since conversion tables from STI to PI exist, one might ask why one cannot just take the data from the STICis module, and directly convert them to PI. The answer is that Intelligibility and Privacy utilize very different levels of the speech modulations. There simply is not enough data in a STICis test to detect the tiny amount of information in a PI test. Accordingly the PI firmware is required to collect the data at the very low levels found in PI. This is also why a PI test takes approximately twice the time of a STICis test. A voice evacuation system would typically be at 75dBA or more. Whereas information being masked for Privacy might well be at levels far below 40dBA. The table showing the relationships is below, with the relative ratings.

STI/CIS/PI Score Card

DOWNLOAD
pdf Download in Adobe® PDF format.
You will need a copy of the Adobe® Acrobat® Reader to view it.

STI Product Information Brochure. 639K.
DSP2 and Talkbox Information in Japanese . 171K.
Understanding Speech Intelligibility White Paper. 223K. This white paper is provided through the courtesy of Bose® Professional Systems.
DSP30 and Test Results White Paper. 49K. This white paper is provided through the courtesy of Bose® Professional Systems and TNO Human Factors.
The Measurement of Speech Intelligibility White Paper. 203K. This white paper is provided through the courtesy of TNO Human Factors.
DSP30 approval letter. 15K.

top Home Pro Audio Page Analyzer Page     Sales Tech Support     Check Price Place Secure Order