STI Program Information

 Version 3.1 - Upgraded 9/9/03

Graphical User Interface (GUI):  

A modular GUI has been designed to improve access to the various tools available within the program.  The Main DialogIDH_MAINDLG provides buttons for launching each component of the STI program.  Components include the Level DialogIDH_LEVELS used to adjust recording levels, the Measurement DialogIDH_CONFIGDLG used to configure recording options, and the Analysis DialogIDH_ANALYZDLG used to configure an STI response analysis.  

Additionally, measurement and analysis progress dialogs now include a cancel button allowing the procedures to be cancelled at any time.  

Measurement Stimulus:  

The stimulus signal in Version 2.0 of the program was modulated white noise.  During an STI analysis, speech weighting was applied to the response before calculating the STI.  A more accurate method, employed in Version 3.1, is to speech weight the stimulus and remove the weighting from the analysis.  This approach provides more accurate results in circumstances where the system under test has limited bandwidth or exhibits non-linear characteristics such as compression.

The stimulus signal in Version 3.1 is modulated ANSI speech shaped noise [8]IDH_REFERENCES with dual tone multiple frequency (DTMF) framing markers used for locating the start and stop position of a response within the recorded WAV file.  The STI analysis depends on precise timing of the modulated segments and these framing markers allow responses to be recorded on external equipment, such as a digital audio tape (DAT) machine, and re-recorded onto the program PC without loss of crucial timing information.  

Open Loop Response Analysis:  

In Version 2.0, immediately after the stimulus was presented the STI analysis procedure started.  This meant that responses had to be recorded through the program PC’s line input.  In Version 3.1 however, responses can now be recorded on external equipment and analyzed after being recorded onto the program PC.  This approach offers much greater flexibility in terms of equipment setup (e.g. recording equipment can be located remotely from the program PC).  

Additionally, multiple stimulus sequences can be presented sequentially so that a large number of responses can be recorded during one measurement session.  

System Requirements  

Minimum Configuration:  

Microsoft Windowsâ 95, 98, NT 4.0, or higher

90 Mhz (or faster) Intel Pentiumâ or compatible processor

8 Megabytes system RAM

25 Megabytes hard disk space

Windows compatible sound hardware with line level inputs and outputs  

About Sound Hardware:

RERC-HE's STI Program is designed to work with most Windows compatible sound cards.  Sound hardware should support full duplex (simultaneous record and playback) operation if local PC response recording is desired.  While most integrated devices and plug-in sound cards

support this feature certain devices may not perform properly due to the inability of device drivers to maintain the data throughput required by the STI program.  If the STI program does not play the stimulus without 'breaking up', or the program issues an "MCI Error", try recording the response on external equipment.  

Recorded Response File Formats:  

All audio files used by the STI program must be mono WAV format, and recorded at a sample rate of 44.1Khz with 16 bits of resolution.  If response files do not conform to this format they will not be analyzed.  

STI Program Files  

Files included in the distribution: (These files must reside in the same directory)  

 Files created by the STI Program: (These files are created in the STI program directory with the exception of response.wav and analysis.log which are created in the analysis directory specified under OptionsIDH_OPTIONDLG.)  

 Equipment Setup  

The STI Program is capable of measuring the speech transmission index of many different types of systems.  While these systems differ in their end application, setup for performing an STI measurement is usually quite similar.  

Typical systems tested with the STI program:  

As is true with any of the systems listed above, all equipment that appears between the stimulus and response WAV files will affect the resulting STI, including the program PC’s digital to analog, and analog to digital converters.  This is because the analysis procedure has been calibrated to yield a perfect STI (1.0) for the stimulus WAV file, stim.wav.  

In the case where the system response is to be recorded on the program PC, two examples of equipment setup are provided below.  If external equipment is used for recording a response, that response must first be re-recorded on the program PC using the WAV formatIDH_SYSREQ and then analyzed using the program’s Analysis DialogIDH_ANALYZDLG.  

Equipment Setup Examples:   

Figure 1: A Loudspeaker or Microphone STI Measurement

Figure 2: An FM System STI Measurement  

NOTE:  

All WAV files used by the STI Program are mono format.  Stereo sound cards play mono files on both the left and right (TIP and RING) channels.  Equipment used to present the stimulus may be connected to either the TIP or RING segments of the LINE OUT jack.  Sound cards record mono files from the left channel only so external equipment used to record the response must be connected to the TIP segment of the LINE IN jack.  Equipment setup should be verified by Adjusting the Recording LevelIDH_LEVELS as described in this help file.  

Main Dialog  

When the STI Program is started the Main Dialog appears.  This dialog window allows the user to select from three procedures:  

1.       Adjust the Recording LevelIDH_LEVELS  

2.       Configure and Perform an STI MeasurementIDH_CONFIGDLG  

3.        Configure and Perform an STI Response AnalysisIDH_ANALYZDLG  

Refer to the respective sections within this help file for more details.  

Additionally, the OptionsIDH_OPTIONDLG button allows the user to set the path to the analysis directory.  

Adjusting the Recording Level  

Pressing the Level button on the Main Dialog toggles on, and off, a small playback control that presents a test stimulus used for monitoring the recording level on the program PC, or external equipment.  The test stimulus is a 10 second, un-modulated ANSI speech shaped noise, with start and stop DTMF framing markers.  Setting the equipment recording level using the test signal will ensure that measurement recordings are of sufficient amplitude for later analysis, and that the recorded responses do not overload external recording equipment during stimulus presentation.  The procedures outlined below are also a quick way of verifying proper equipment setupIDH_EQUIPMENT.  

Recording on the Program PC:  

1.       Open the Window’s Volume Control [Start > Programs > Accessories > Entertainment > Volume Control] and select Options > Properties > Adjust Volume for > Playback.

2.       Make sure the Wave Out device is not muted and that its volume slider is set to the maximum setting.

3.       On the Volume Control select Options > Properties > Adjust Volume for > Recording.

4.       Make sure the Line Input has been selected. Adjust the Line volume slider to approximately 75% of the maximum setting.

5.       Make sure that the system under test has been connected to the PC’s Line Input and Line Output as described under Equipment Setup IDH EQUIPMENT.

6.       Open the Window’s Sound Recorder [Start > Programs > Accessories > Entertainment > Sound Recorder].

7.       Start the STI program and press the Level button on the Main Dialog.  This opens a small playback control.  If necessary, move the playback control, Volume Control, and Sound Recorder so that all are visible.

8.       Press “record” on the Sound Recorder and begin playback of the test stimulus by pressing the play button on the playback control.

9.        While observing the recording level meter on the Volume Control’s Line input, adjust either the Volume Control’s Line slider, or external equipment levels, so that the level meter registers as high as possible without clipping.  Ideally, the level meter should register within the highest yellow band of the meter and very infrequently peak into the first red band.  

Recording on External Equipment:  

1.       Open the Window’s Volume Control [Start > Programs > Accessories > Entertainment > Volume Control] and select Options > Properties > Adjust Volume for > Playback.

2.       Make sure the Wave Out device is not muted and that its volume slider is set to the maximum setting.

3.       Make sure that the presentation side of the external equipment is connected to the PC’s Line Output and the response side of the external equipment is connected to an appropriate recording device (e.g. DAT or another PC).

4.       Start the STI program and press the Level button on the Main Dialog.  This opens a small playback control.

5.       Begin recording on the external equipment so that its VU level meters are activated and then begin the test stimulus output by pressing the play button on the playback control.

6.        While observing the VU level meter on the external equipment, adjust external equipment levels, so that the VU meter registers as high as possible without clipping.  

Options  

In order to simplify organization of analysis results, the STI program offers the option of specifying a different directory in which to store response files and analysis reports.  This directory should be created before setting it as the Analysis Path.  

To set the Analysis Path:  

1. Press the Options button on the Main Dialog

2. Enter the desired directory path in the space provided (Analysis Path).

3. Press the Save button.

NOTE:

The directory entered must be a valid pathname and should not include a trailing backslash.  Pressing the Reset button sets the Analysis Path to the same directory as the Program Path.  

Measurement Dialog  

Pressing the Measurement button on the Main Dialog opens the window used to configure an STI measurement.  The STI measurement itself consists of playing a specially designed stimulus signal through the “transmission” side of the system under test while simultaneously recording the system’s response to that stimulus.  Configuring an STI measurement only requires selecting from the options listed below and pressing the OK button.  

Recording Options:  

1. This PC:  The program PC’s Line Input is recorded at the same time the stimulus is presented to the program PC’s Line Output.  The recorded response is saved to the specified Response Filename.  The default name for the file is response.wav however any valid name up to 8 characters in length can be specified.  The .wav extension is added automatically by the program.  This recording option requires that the PC’s sound hardware support full duplex operation described in System Requirements.IDH_SYSREQ

2. External Equipment:  The system response is recorded on external equipment and the PC’s Line Input is ignored.  The system response must then be recorded to the program PC and analyzed to obtain an STI for the measurement.  Full duplex operation of the sound hardware is not required when using this option.

• Number of Trials:  For either of the recording methods listed above, a sequential number of stimulii may be presented.  Number of Trials is limited to between 1 and 99.  When recording on the PC, a single response file is saved.  When recording on external equipment there is no restriction that the sequential responses be recorded to the program PC as a single WAV file.

Logging Options:  

·         Measurement Log File:  Enter a filename in the space provided.  Measurement success, or failure, will be recorded here in text format.  The default filename is measure.log but any valid filename up to 8 characters can be specified.  The .log extension is added automatically by the program.  

·         Append:  Un-checking this box will cause the specified logfile, if present, to be overwritten.  If the specified logfile does not exist this checkbox has no effect.  

·         Comment:  Enter information regarding conditions of the test here.  This comment will be saved in the logfile and can be used in identifying the measurement pass.

NOTE:   

Due to the length of the stimulus signal it is important to keep track of available hard disk space.  A recording of 99 sequential trials will consume as much as 1 Gigabyte of hard disk space.  

AN EXAMPLE LOG FILE:  

********************************************************************************

Lexington Center and School for the Deaf

Speech Transmission Index Program

Version 3.1

Measurement Log File

********************************************************************************

Date:                Thu Mar 27 09:31:34 2003

Stimulus File:            stim.wav

Response File:            test1.wav

Comment:            Radio Shack SPL Ref 80

Trials:               1

--------------------------------------------------------------------------------

Presentation Successful

********************************************************************************

Date:                Thu Mar 27 09:34:01 2003

Stimulus File:            stim.wav

Response File:            test2.wav

Comment:            Radio Shack SPL Ref 90

Trials:               3

--------------------------------------------------------------------------------

Presentation Successful  

Analysis Dialog  

Pressing the Analysis button on the Main Dialog opens the window used for configuring an STI response analysis.  The STI response analysis consists of two phases; The Marker Detection Phase, and the STI Calculation Phase.  Before an STI calculation can be performed, the Marker Detection Phase determines if any valid STI stimulus responses are present in the selected WAV file(s).  This procedure uses dual tone multiple frequency (DTMF) markers present in the stimulus, to detect the start, and stop, position of a response.  If no valid responses are detected the WAV file is closed and the analysis is aborted.  If valid responses are detected the STI Calculation Phase is started.  To run an STI analysis simply enter the options listed and press the OK button.  

• Analysis Log File:  Enter a filename in the space provided.  Analysis results will be recorded here in text format.  The default filename is analysis.log but any valid filename up to 8 characters can be specified.  The .log extension is added automatically by the program.

·         Append:  Un-checking this box will cause the specified logfile, if present, to be overwritten.  If the specified logfile does not exist this checkbox has no effect.  

• Select File to Analyze:  This box will list all WAV files present in the current analysis directoryIDH_OPTIONDLG.  Select any file from the list and press OK.  An analysis will be performed on the selected WAV file.

• Batch Process All:  Check this box to analyze all WAV files from the list.

AN EXAMPLE LOG FILE:  

********************************************************************************

Lexington Center and School for the Deaf

Speech Transmission Index Program

Version 3.1

Response Analysis Log File

********************************************************************************

WAV File:                     noinp.wav

Recorded:                     Wed Mar 26 14:06:16 2003

Analyzed:                      Wed Mar 26 14:06:53 2003

Total Responses:         0

--------------------------------------------------------------------------------

End of WAV File Analysis

********************************************************************************

WAV File:                     response.wav

Recorded:                     Wed Mar 26 14:03:44 2003

Analyzed:                      Thu Mar 27 09:16:52 2003

Total Responses:         1

--------------------------------------------------------------------------------

Response:1            STI =            0.86

Average SNR (dB) in Octave Bands 125-8000 Hz

   -7.89   10.96   11.95   13.82   14.24   14.51   14.75

--------------------------------------------------------------------------------

End of WAV File Analysis  

Theory behind the Speech Transmission Index  

The RERC’s Speech Transmission Index Program is based on the Modulation Transfer Function (MTF).  The MTF, traditionally used in the study of room acoustics, has been successfully applied to the measurement of speech intelligibility [1]IDH_REFERENCES.  The MTF is a measure of the reduction in modulation of a test signal due to additive noise and temporal, or non-linear, distortion resulting from transmission through the system under test.  Figure 1. shows a graphical representation of the MTF measurement process.  A more detailed description of the MTF can be found in [3] IDH_REFERENCES and [4] IDH_REFERENCES.  

The stimulus signal used by the STI program consists of ANSI speech shaped noise modulated by low frequency sine waves.  Fourteen modulation frequencies are used in the STI stimulus.  The modulating sine waves range in frequency from 0.63Hz to 12.7Hz in 1/3 octave steps.  The stimulus is designed to simulate the intensity distributions found in running speech.  The upper portion of Figure 1 shows one segment of the stimulus and the response of a typical system.  Note that the response has been corrupted by additive noise.  Other sources of corruption may include temporal (blurring), or non-linear (clipping) distortion.  This corruption results in a response with lower modulation factors than the original stimulus.  The MTF is a measure of the amount of modulation reduction versus the modulation frequency.  

Figure 1.  Calculating the Modulation Transfer Function  

The STI program calculates modulation reduction factors for seven one-octave spectral bands important to the transmission of speech.  These bands range from 125Hz to 8000Hz in one octave steps.  Within each of the seven octave bands, fourteen modulation reduction factors are calculated resulting in a matrix of 98 total factors.  The modulation reduction factors are converted to effective signal to noise ratios (SNR).  These SNR’s are weighted and averaged to obtain the resulting STI.  A detailed description of the SNR conversion, weighting, and averaging process can be found in [3] IDH_REFERENCES.  

References  

[1]  Steeneken, H.J.M. and Houtgast, T. (1973).  “The Modulation Transfer Function in Room Acoustics as a Predictor of Speech Intelligibility”, Acustica 28, 66-73 (1973).  

[2]  Steeneken, H.J.M. and Houtgast, T. (1979).  “A physical method for measuring speech transmission quality”, Institute for Perception TNO, Soesterberg, the Netherlands.  

[3]  Steeneken, H.J.M. and Houtgast, T. (1984).  “A review of the MTF concept in room acoustics and its use for estimating speech intelligibility in auditoria”, Institute for Perception TNO, Soesterberg, the Netherlands.  

[4]  Studebaker, G.A. and Matesich, J.S. (1992).  “A derivation of the Modulation Transfer Function”, Memphis State University, Memphis, Tennessee.  

[5]  Rife, D.D. (1992).  “Modulation Transfer Function Measurement with Maximum Length Sequences”, DRA Laboratories, Sterling, Virginia.  

[6]  Banks. K. (2002).  “The Goertzel Algorithm”, Embedded Systems Programming, CMP Media LLC, Manhasset, New York.  

[7]  Schmer, G. (2000).  “DTMF Tone Generation and Detection:  An Implementation Using the TMS320C54x”, Document Number SPRA096A, Texas Instruments, Houston, Texas.  

[8] ANSI (1992).  “Testing Hearing Aids with a Broad-band Noise Signal”, American National Standards Institute, ANSI S3.42-1992.  

The RERC’s STI Program and associated documentation written by:  

Michael Steele
Senior Research Engineer
c/o City University of New York
Graduate School
365 Fifth Ave.
New York, NY  10016-4309
steelejam@aol.com

Technical Support  

Prior to contacting the RERC's Technical Support Team please review the instructions found in the System Requirements IDH_SYSREQ, Equipment Setup IDH_EQUIPMENT, and Adjusting the Recording Level IDH_LEVELS sections of this help file.  If problems persist contact Technical Support at the number listed below.  

Matt Bakke, Ph.D.
Director, RERC on Hearing Enhancement
Gallaudet University
800 Florida Ave., NE
Washington, DC  20002
202-651-5335 v/t
matthew.bakke@gallaudet.edu