Brüel Acoustics


Rasti 44BA


Acoustic Ouality in Rooms

Before W.G. Sabine around 1900 defined the reverberation time T60 acoustic quality was a great mystery.
Nobody could understand why some rooms had a good acoustic and others, similar looking, had an inferior acoustic.
A few things were observed: Bare hard walls in rectangular rooms were nearly always bad, but not always.
Rooms with irregular surfaces, filled with sculptures, chubby- cheeked cherubs, wooden furnitures and some tapesty were often good, but not always .
So building a hall for both plays and concerts was an adventure, where some luck was needed.

After the reverberation time was introduced the worse mistakes were rare, but still some halls with equal reverberation time could have very different acoustic quality.
Another problem was that T60 often was very different.

B&K became famous for in 1945 in Sweden to have developed a unique level recorder which was used for accurate T60measurements.
B&K had in fact a world monopoly for level recorders for more than 40 years.
No competitors could match the principle, quality, and price. B&K made more than 25.000 level recorders during the years.

Now T60 is measured with a PC for a fraction of the former cost.
With the level recorder it was possible to find out that T60 depended on where in the room it was measured.
According to the simple Sabine theory T60, should be the same anywhere in the room.

The reason for this apparent anormality, we now know, is that the reverberation field is built up of many different resonances, eigentones wandering around in the room with different damping.
Especially the one-dimension eigentones like flutterechos and two-dimensional eigentones have a less damping than the three- dimensional eigentones causing a reverberation curve not correct exponential.

Another more direct indicator for acoustic quality is Speech Transmission Index, where the percentage of correctly understood syllables is indicated.
Dr. Steeneken in Holland have developed in 1973 a more simple method called Rapid Speech Transmission Index -RASTI.
This method is now approved by IEC as Rec.268-16.

As RASTI is a new and excellent method for evaluating acoustics in rooms B&K became interested and with assistance from dr. Steeneken constructed RASTI-meter 3361.
This instrument was the first and is still the only commercial accessible RASTI in the world.
Unfortunately too few instruments were sold with the result that B&K did not follow up on the construction.
To-day the technique in the B&K design is more than 25 years old and accordingly very expensive to produce.

Brüel Acoustics BA could see the need for a good inexpensive STI-instrument and devlopped for its own use some instruments to support the sale of AcoustiCone.

Brüel Acoustics devlopped 44BA wich have some unique and new features, and is ideal for evaluation of room acoustic.

FOR GOOD ACOUSTIC: USE STI.

RASTI TRANSMITTER (pdf 40 KB)

RASTI RECEIVER (pdf 37 KB)

Measurements at Opera Theatre of Roma (69 Kb)
Rasti 44BA (38 Kb)


STI- RASTI

Rapid Speech Transmission - 44BA

Knowing the 25 years old B&K system 3361, Brüel Acoustics decided to develop a modern STI system and include some new and original thinking, which are as follows:

1) The instrument is formed as a human head and torso, which produce a correct directivity pattern.
(B&K with 2 KHz 10 dB).
2) The head is a sphere and a monopole can be made using very little material. (B&K: Rectangular heavy box).
3) Both power and battery operated. (B&K only battery).
4) Three levels: Standard, +10, and +20 dB. (B&K only +10 dB).
5) Both manual, computer and radio controlled. (B&K manual).
6) Receiver is an interface to PC which can be used to make highlevel programmes. (B&K special unit excellent, but heavy and expensive).
7) Can use B&K microphone, and other brands.
8) The whole instrument packed in a lightweight soft lined transport.
9) The instrument is patent protected under Danish patent. (B&K has no protection).

The general conclusion is that 44BA could be sold in large quantities by Brüel Acoustics BA.
There is a heavy pressure from customers to have a good and economical STI-system.


The RASTI method for the objective rating of speech intelligibility

1. General
The RASTI-method is an objective method for rating the transmission quality of speech with respect to intelligibility.

The RASTI-method is a condensed version of the STI-method (see reference 1), based upon the measurement of the Modulation Transfer Function. The method is intended for rating speech transmission in auditoria, halls and rooms with or without sound systems.

The Rapid Speech Transmission Index (RASTI) method has been incorporated into an instrument that quickly measures the Speech Transmission Index. It allows to evaluate the quality of speech intelligibility in rooms, lecture halls, theaters, churches, classrooms, auditorium, etc.

This instrument replaces the laborious time consuming method of using a speaker and a listener in various positions to evaluate the Speech Transmission Index. The system developed by Brüel Acoustics can correctly evaluate the Speech Transmission Index for more than fifteen languages.

It is economical and time saving, since each station can be evaluated in eight, sixteen or thirty-two seconds. It allows a more thorough and accurate evaluation of the acoustical properties of a room at a decreased cost in manpower. This becomes very important in improving the acoustics of a room since it allows the evaluation of various placements of absorption material. Thus, the best acoustical results can be obtained for the amount of absorption material used, minimizing costs.

2. Description of the method
The STI-RASTI-method is based on an acoustic test signal, produced at a talker?s position and received by a microphone at a listener?s position. The measurement parameter is the reduction of the modulation index of the test signal. The reduction is interpreted in terms of an apparent signal-to-noise ratio. The result is independent of the cause of the reduction, which can be reverberation, echoes, interfering noise or some combination of these.

The acoustic test signal is produced by a sound source simulating the acoustical characteristics of a real talker. The important characteristics of the sound source are:

- physical size
- directivity
- position
- sound pressure level

The sound source shall be mounted in an appropriate enclosure, with dimensions of the same order as the human head. The directivity index (for the reference axis at 0 azimuth and 0 altitude) shall be 1-3 dB at 500 Hz and 2-5 dB at 2 kHz. The frequency response measured in any direction shall be reasonably flat within each of the octave bands centered at 500 Hz and 2 kHz. The frequency response measured on the reference axis shall be flat within 2 dB in each octave band. The sound source shall be situated at the position of a real talker?s mouth. The reference axis shall be pointed in the expected speaking direction of an actual talker in the auditorium.

As microphone positions any position which could be occupied by a listener in the auditorium can be used. The measuring microphone shall be omi-directional.

3. PROCEDURE
3.1 The detailed calculations

The complete RASTI measuring procedure consists of the following steps:

a) Specify the equivalent A-weighted sound pressure level (Leq, A) of running speech representing the talker. This speech level, serving as reference level, may be determined by measurements on the performance of actual talkers in the auditorium under test at a distance of 1 meter in front of the talkers? mouth. In the absence of such measurements, a value of Leq, A = 60 dB is recommended as reference level.

The test signal is adjusted so that the long-term RMS level for the octave band with center frequency of 500 Hz is ?1 dB relative to this reference. Similarily, the long-term RMS level for the 2 kHz octave band should be adjusted to a level 10 dB lower than this reference. All level measurements shall be done on the reference axis of the sound source at a distance of 1 meter.

b) Apply sinusoidal intensity modulation, for the 500-Hz octave band at modulation frequencies of 1,2,4, and 8 Hz and for the 2 kHz octave band of 0.7, 1.4, 2.8, 5.6 and 11.2 Hz (all modulation frequencies have a tolerance of plus and minus 0.5%).

c) Apply octave filtering at center frequencies of 500 Hz and 2 kHz to the microphone output and analyze the two filter-output signals for deriving the modulation index for each relevant modulation frequency in the resulting intensity envelope.

d) Determine the modulation reduction factor m (the ratio between the resulting modulation index and the initial modulation index of the test signal), of the relevant modulation frequencies in both octave bands (m < 1).

e) Convert each of the nine m-values into a value accoring to:

x = 10 log [m / (1-m)] (1)

This may be interpreted as an apparent signal-to-noise ratio in decibels.

f) Truncate the obtained values of when exceeding the range of plus or minus15 dB.

g) Determine the arithmetic mean x of the nine values thus obtained.

h) Normalize to an index y, ranging from 0 to 1:

y = ( + 15) / 30 (2)

This is the RASTI-index for the chosen conditions. (If not otherwise stated it is assumed that the reference speech level is 60 dB(A)).

3.2 Precision of the RASTI measuring procedure

Because the test signal is band-limited random noise, repetition of a measurement will not normally produce identical results, even under conditions of steady interference. Rather, the results will center around a mean with a certain standard deviation. This may depend, among other things, on the measuring time involved. An example of the RASTI procedure is shown below.

3.3 Limitations of the RASTI-method

The absolute accuracy of the RASTI-method is affected by several factors such as unusual background noise, distortion, widely changing reverberation with frequency, etc.. Therefore, the application for absolute measurements should be restricted to cases where the following requirements are met:

a) Essentially linear speech transmission (no clipping etc.), as the RASTI-method does note account for non-linear distortions.

b) Wide-band speech transmission (typically 200 Hz ? 6kHz), as the RASTI-method is based on the assumption of an essentially unlimited speech spectrum.

c) Conditions where the background noise does not contain audible tones, nor shows marked peaks or troughs in the octave-band spectrum.

d) Conditions where the background noise is not of an impulsive character.

e) Conditions where the reverberation time is not strongly depending on frequency.

If all of the above requirements are not met, the results should only be used for comparisn purposes for measurements made under essentially the same conditions.

3.4 Applications

The RASTI-index can primarily be used for comparing speech transmission quality at various positions and for various conditions within the same auditorium. In particular, it can be used for assessing the effect of changes in the acoustical properties of the auditorium, including the effects of the presence of an audience, or of changes in any sound system installed in the auditorium (see reference 3).

The RASTI-index may also be interpreted as an absolute rating of speech transmission quality with respect to intelligibility when comparing different auditoria under similar conditions. An example of such an interpretation is given in Appendix B.

Note: In cases where one or more of the requirements of Clause 3.4 are not met, it may be advantageous to use the STI-method (see reference 1).

References:

1. Steeneken, H.J.M., and Houtgast, T., A physical method for measuring speech transmission quality, J.Acoust., Soc., Amer. 67 (1980), 318.

2. Houtgast, T., and Steeneken , H.J.M., Report on a multi-language evaluation of the Rapid Speech Transmission Index method for estimating speech intelligibility in auditoria. Report IZF 1981-12, Institute for Perception TNO, Kampweg 5, Soesterberg, The Netherlands.

3. Steeneken, H.J.M., and Houtgast, T., Some applications of the Speech Transmission Index (STI) in auditoria. Accepted for publication by Acustica, 1982.

4 EXAMPLES OF ACOUSTICAL SOLUTIONS

Using the RASTI method, the STI is rapidly evaluated in different areas of a room by the RASTI-index.

The software allows, the user to record a description of the room analyzed as well as the evaluation of the RASTI-index at various locations.

Fig. 1. Classroom 317 as it is used for teaching without special attenuation to the acoustics. This classroom is typical of many classrooms. The average RASTI index for three measurement values at 9 locations is 0.54 with a standard deviation of 0.02

So how can this classroom be improved? One obvious method is to reduce the reverberation time by installing absorptive materials. Brüel Acoustics will sell such material in late 1999. These will be the Acousti Cones and Acousti Cassini. The following example using these devices, shows the usefulness of the RASTI method for determining the best location for placing the absorbing material. This example also demonstrates that just bringing the reverberation time down may not be the optimum use of materials.

Consider two cases in which the same amount absorption material is used for the same room, but placed differently.

Case 1, the absorption material placed in the center of the room.

Case 2, the absorption material is placed so that part of the ceiling is used as a mirror with superior results. Both rooms have the same reverberation time, 1.1s.

RASTI Value for case 2.

Detail analysis at various locations in a room using the RASTI method.

Note: That for the same amount of absorption material, case2 provides better speech intelligibility.

In summary

1 The location of absorbing material is important if the best acoustical results are to be obtained.

2 The RASTI method, with a minimum of effort, allows the user to optimize the location of sound absorbers in order to obtain the maximum STI.




® 1995-2003 All right reserved - Web masters Mario & Paolo Mattia -
- updated on 2003, February

On line from 1995 November