rasti applications

8/18/2019 Rasti Applications

1/11


2/11


3/11

RASTI measurements:Demonstration of different applications

by Thomas R. Horrall, BBN, Cambridge, M.A.

Torben Jacobsen, Bruel & Kjasr, Denmark

Introduction

Speech intelligibility in thea tre s, veloped. RAST I, which sta nds for sion Meter, Tra nsm itt er Type 4225lecture rooms, churches , airpla nes, etc. Rapid Speech Trans miss ion Index, is and Receiver Type 4419.is an impo rta nt consider ation, which a new stan dardi zed meth od.is why its evaluati on is often require d. The RAS TI meth od and the foun-Subjective met hods can be used for The aim of this applic ation note is dation for it is briefly ment ione d be-this evaluation, in which people assess to describe and demons tra te special low and described in detail in refer-the intel ligibi lity at different places in uses of the RASTI met hod . The work ences [1 - 5] and is being st and ard izedthe room. This meth od is rat her time was performed in cooperati on between by IEC [6]. For anot her use of theconsuming and th e devi ation of th e Bruel & Kjser and Bolt, Ber anek and RAS TI met hod see [7].results is rath er high. For these rea- Newma n, Cambridge . The inst rume n-sons objective met hods have been de- tat ion used was the Speech Tra nsmi s-

The RASTI Method

The basis for the RASTI method isa measurement of the modulationtransfer function between the transmitter (the speaker) and the receiver(the listener). Conversion of the modulation transfer function to a speech

Fig. 2. The Speech Transmission Meter, Receiver Type 4419. The signal ismeasured with the microphone, butalso direct electrical input is possible. The measurement time can be 8,

Fig. 1. Front and rear panels of the Speech Transmission Meter, Transmitter Type 4225. 16 or 32 seconds. After the measure-The Ref. Output level is according to the IEC standard and is equivalent to a speech ment the RASTI value is displayed.level of 60 dB(A) (Leq) at 1 m from the speaker. Three different output facilities exist: Other results can be called to thea built-in loudspeaker, an output for external speaker and an electrical output. The display by means of the appropriate

box together with the built-in loudspeaker have approximately the same directivity pushkeys or read out by means ofas that of a human being. the serial interface.

3


4/11

intelligibi lity rating was developed by method it is possible to evaluate the listener' s position and by means of athe Insti tute for Perception, TNO, the intelligibility in one listening position discrete Fourier Transform the modu- Netherlands . This rat ing is called the in 8 seconds. lat ion transfer funct ion is measuredSpeech Transmission Index, STL and from this the RASTI value is cal-

During a measurement the t rans - culated. The result appears in the Re-The disadvantage of the full STI mit ter (see Fig. 1) is placed at the ceiver display and can be printed out

method is that it demands 98 mea- speaker's position and sends out a test on a printe r. Besides the RASTI val-surements, and therefore a quicker signal which contains information ue, other information of diagnosticmethod - the RASTI method - has about the speech spectrum, fluctua- value is provided. This facility ensures

been developed and is now being stan- tions in speech, and speech level. The th at the RASTI instruments also candardized. By means of the RASTI Receiver (see Fig. 2) is placed at the be used as a diagnost ic tool.

Sound Masking System in Open-plan Offices

Studies of office workers' satisfac- background noise level is a critical corresponding to mouth height for ation over the last 20 years have dem- part of assuring that the masking sys- seated speaker (see Fig. 3). Theonstrated the importance of obtaining tern is subjectively unobtrusive , and RASTI receiver was located in the ad-an adequately low level of speech in- accepted as par t of the normal envi- jacent office (HI RTLE) and similarlytelligibili ty between adjacent offices. ronment by the office worker. positioned corresponding to ear heightMost workers are dis tracted and find for a seated listener. The height of theit difficult to concentrate if they can A standard sound level meter with par tit ion dividing the two offices wasunders tand unwanted and intruding A-weighted scale can be used to indi- 62" = 1,57 m, and all ceiling and par ti-conversations . cate the perceived loudness of back- tion surfaces were well treated with

ground noise, but cannot indicate in- sound absorbing material s. The endSpeech intelligibility depends pri - telligibili ty direct ly. As a resul t, vari- walls of both offices were acoustically

marily on the ratio of sound levels of ous objective and subjective tes t hard .speech to background noise and rever- methods have been developed to esti-

beration experienced by a lis tener. mate inter-office inte lligibi lity. These Separa te sets of RASTI measure-However, good open-plan office con- methods typically requi re a skilled ments were made for the transmi tterstruction assures very low reverbera- tes t operator, are time consuming, and directed toward the receiver, for twotion so that only the speech to noise suffer from limited repeatabili ty and source levels corresponding to normalratio is import ant . Good office con- reliabil ity. Clearly, an ins tru men t ca- speech and raised voice, and for threestruction also reduces speech sound pable of making rapid and reliable adjustments of background sound lev-transmi tted from one office to another measurements correlated directly with el.

so that a comfortable level of back- intelligibility would be a valuable tool.ground noise will assure that the intel - To evaluate the RASTI method for The background sound level wasligibility is no higher than about 0,20 determin ing intelligibility, in situ measured in octave bands with a pre-as measured by the arti cula tion index measurements were made in open- cision sound level meter /fil ter set, and(AI), ANSI standa rd S.3-2. plan offices of BBN Laborato ries was also confirmed in two bands with

which are equipped with a calibrated the RASTI receiver, see Table 1.Open-plan offices are extremely and adjustable sound masking system.

popular in the USA. Quality offices of The octave band sound isolation be-this type are normally provided with The RASTI transmi tte r was located tween offices had been measured ear-extensive acoustical treat men t of ceil- in one office (PIRN) and positioned lier for the same tra nsmit ter and re-ings and partial-height dividing partitions in order to reduce transmittedspeech sound. However, the background noise from conventional

sources, such as the building ventilating system, is normally insufficient toassure low enough intelligibility. As aresult, electronically generated andfrequency contoured sound provided by a sound masking system has become a part of most high quality open- plan office ins tal lat ions.

The ideal adjustment of a soundmasking system is that which has botha spectrum shape and sound levelyielding the lowest perceived loudness, and which simultaneously pro

vides the 0,20 AI or other intelligibility goal. Achieving the lowest possible Fig. 3. A plan of the open-plan office

4


5/11

ceiver positions, using broa dban d ran- I Resultsdom noise played th rou gh a small 1 , , . .louds peake r wit h di rec tivit y approxi - Octave band frequency 250 | 500 | 1000 | 2000 | 4000 Hz

mat ing th at of the hu ma n voice and Background Noise

with a preci sion sound level me te r/ ^ BBN normal spac e average AQ I _ I 7 7 H ~~1 QO HDfilter set. The se measur eme nt s al- design levellowed direct calculation of the Al, us-

T-JT^T , 2. Measured at normaling a BBN compu ter progra m . . . . , 0 .n _. ._ ._ . . _„ ._ ,J? r .-KTX /. ■ • i i masking level + 3 dB, 51 49 46 44 36 dB(OPLAN), for comparison with those S L M a t HIRTLE

derived from RASTI measurements.Translation from RASTI value to AI 3 ' Measured at normal level,. , jn ^ i i i M I + 3 dB, RASTI receiver - 49,0 - 43,7

is according to the method described HIRTLE - 4 9 1 - 44 3in Acusti ca, Vol. 46, 1980, p.60. For - - pu re noise in te rf erence (no reve rber a- 4- N o r m a l l e v e ! f r o m 2


6/11

meas urements were made and the cult) for the opera tors to communi- In other connect ions, PA systemsRASTI results were 0,45, 0,47 and cate, but only if they stand close to have been evalua ted in power plan ts0,47, indica ting "fair" intelligibility. each other and speak loudly. with backgr ound noise levels of

90-100 dB(A) and where the reverber-The resul ts show that it is possible It can be concluded th at the tele- ation time has a large influence. In

to make RASTI measurements in a phone in the factory hall is not of these applications the RASTI methodfactory in spite of high backg round much use, but the telep hone in the has also been shown to be valuable .noise. Fur the rmo re , the resul ts in this office can be used although it is stillcase show that it is possible (but diffi- necessary to speak loudly.

Concert/Drama Hall with Adjustable Reverberation Time

Fig. 5. The seating plan for the concert/drama hall

51 Walden St reet is a smal l aud ito - I RASTI Values at Seating Location Numberrium used both as a concert hall and a | ~ | ~ l ~ l ~ l ~ | ~ l ~ I ~ 1 ~ I ~drama theatre by community per-forming arts groups in Concord, Mas- s 'de Wall Panelssachus etts. It has acoustical panels Soft,Ceiling Q 5? Q 5g Q 5 4 Q g1 Q 54 Q 5 6 Q 55 Q 55 Q 5 9 Q 5 7

i j-u -J ii u- u u A Pane s Hard Sidealong the side wall which can be ad- _

justed for sound absorp tion to lowerthe reverberation time for theatrical S i de Wa"_ P a n e l s

use, or adjusted for sound reflection to p a r d ' C^ l h l l 9 0,54 0,51 0,51 0,49 0,48 0,57 0,54 0,51 0,57 0,53increase the reverberation time and S j d D o w nearly latera l energy fraction for musi- —;cal performances. Opposite ends of " e Wa", P a n e l s

the hall are used as the stage for musi- n a r d ' , C | l h « 9 o^ - °' 5 0 °' 5 1 °- 54 ° ' 5 3, , , ° A. ,T Panels Soft Side

cal and drama presen tati ons. No Down

sound ampl ification is used for either ' ' ' ' ' ' ' ' ' ' T OO813GBC^^ ^ ' Table 3. Results from the different positions shown in Fig.5 when the hall is used as theatre

7

RASTI Values at Seating Location Number1 2 3 4 5 6 7 8 9 10

Side Wall Panels

Soft, Ceiling

Panels Hard Side

Down

0,57 0,58 0,54 0,51 0,54 0,56 0,55 0,55 0,59 0,57

Side Wall Panels

Hard, Ceiling

Panels Hard

Side Down

0,54 0,51 0,51 0,49 0,48 0,57 0,54 0,51 0,57 0,53

Side Wall Panels

Hard, Ceiling

Panels Soft Side

Down

— — 0,50 0,51 0,54 0,53 — -—

T00813GRn


7/11

In addition to the side-wall acousti- munication for musical performance. The results indicate a substantialcal panels, there are panels above the For these measurements the RASTI improvement in cross-stage communi-music stage which are acoustically ab- transmit ter and receiver were set up cation when the local sound level wassorbing on one side and reflecting on on opposite sides of the musical stage adjusted for 60 dB(A) and the over-the other: these can be inver ted with and music stands were interposed be- head panels were adjusted for soundthe absorbing side down for loud mu- tween them to block the direct line of reflection from one side of the stage tosical groups such as brass bands, or sight to simulate the barrier effect of the other. For the same local soundthe reflecting side down for symphony musicians and inst ruments in the mid- condition, but with the panels absorb-or small musical ensembles. With the die of the stage. ing, the RASTI decreased by more

reflecting side down, cross-stage com- than 0,1. Although 60dB(A) is muchmunication (ensemble) should be en- An importan t aspect of conditions lower than a typical orchestral soundhanced (RASTI value increased) so for a good ensemble is the ratio of level, the ratio of sound from thethat , for instance, the string-bass "wanted" (distant instr ument ) sound RASTI transmi tte r to the local noise players can better hea r the violin to local inst rument sound. In order to source may be more typical. In anysound and vice versa. Using these pan- dete rmine what level of local sound case, this resul t corroborates reportsels with the reflecting side down may could be generated while maintaining from musicians who have played withalso provide ceiling reflections to audi- an acceptable remote/local ratio, a the panels set both ways. Compari sonence seated on the music stage for dra- variable level broadband random of resul ts with similar measurementsma productions at the opposite end of noise generator/amplifier/loudspeaker in other halls could be most useful inthe hall. These reflections could be unit was placed near the RASTI re- determining relative stage ensemble par ticularly useful since these seats ceiver. conditions.are furthest removed from the drama

stage. The RASTI value was then mea- A s e c o n d possible use of RASTI in

sured as a function of the local sound m u s i c h a l l s c o u l ( j be as a measure ofRASTI measurements were under- level as well as with the overhead pan- clarity for various sections of the audi-taken to assess the effect of different els adjusted for cross-stage sound re- e n c e Eit her too much claritv (high panel adjustments. The first set of flection or absorption. The local sound RASTI values) or too lit tle is probablymeasurements was taken for the hall level was measured with an SLM and undesirable Through measurementsused in the theatr ica l configuration. the RASTI receiver, and in one case in a variety of halls it should be possi-(See Fig. 5.) The RASTI transmitter the local sound level was simulated by ^le to establ ish a range of RASTI val-was located on stage with the level set programming the receiver for a part ic- u e s generally considered to representto IEC + 10 dB to simulate a tra ined ular background noise level. The mea- p.Q0C[ hearing conditions.actor voice level. The resul ts at differ- sured results are shown in Tabl e 4.ent seating locations are found in Ta ble 3.

Because the hall is quite reverber- i 1 1 1ant when unoccupied (as measured), Local Sound Level Ceiling Panels 1st row 1st & 2nd row

i T-> A

n m T i i j. • n i all absorbi ng reflectin g reflectingthe RASH values do not vary widely . . r _ zat different locations. However, mea- 5 0 0 H z 2000 Hz RASTI value R A S T | R A S T |surements do indicate that intelligibil- a B A dB/oct. dB/oct. value valueity is enhanced with absorbing sidewalls. In addition , location No.6 at the 40 0,54 0,59 0,56 0,56center rear of the hall provides slightly 5 0 °- 47 ° ' 5 3 °- 52 °- 53

improved hearing conditions with the ®° 51 55 ° ' ^ ° ' ^ ° ' ^ ° '^F , , , 6 „ ,. . , , 65 0,23 0,23 0,26 0,24overhead panels reflecting side down. 7 0 n i n n i c- n i o n i oThese resul ts correspond with the ex- ' 1 1 - 1 - 1 - 1 - perience of the users Noise floor of 4419 used instead of background noise:

60 51 55 0,32The second set of measurements TOOSUGBO

was taken to assess cross-stage com- Table 4. Results from the measurements on the concert stage

Evaluation of sound reinforcement systems for a performing arts center

The Filene Center at Wolf Tr ap acoustical and sound system consul- There is an enclosing roof, and theFarm Park is a large outdoor/ indoor tants for the reconstruction. rear of the main floor and balcony ismusic and theat er facility near Wash- completely open to a sloped lawn so asington, D.C. Recent ly rebui lt after a The main bui lding provides for to provide visual and acoustica l corn-disas trous fire, the facility is used for about 3,500 seats on a main floor and munication to the outdoor audience

both popular and classical music per- a single large balcony. The side walls seated there. The lawn may accommo-

formances of all types. Opera is also consist of alt ernating sound reflecting date upwards of 2,500 audience mem- presen ted. BBN Laboratories were the panels and openings to outdoors. bers .

8

Local Sound LevelCeiling Panels

all absorbing

1st row

reflecting

1st & 2nd row

reflecting

dBA500 Hz

dB/oct.

2000 Hz

dB/oct.

RASTI value

1st 2nd

Measurement

RASTI

value

RASTI

value

40

50

60

65

70

51 55

0,54

0,47

0,29

0,23

0,10

0,59

0,53

0,34

0,23

0,15

0,56

0,52

0,39

0,26

0,13

0,56

0,53

0,41

0,24

0,13

Nois e fl


8/11

Fig. 6. Plan of the performing arts center

Sound amplification systems for the cony overhang. A separate under-bal- identical suspended loudspeaker clus-facility include a large 3-way loud- cony system of dist ributed ceiling- ter s are provided for each of the thr eespeaker cluster above the proscenium mounted loudspeakers serves the sections of the balcony (center andon the hall center-line to serve the audience seated there , most of which each side). The under -balcony and

forward par t of the main floor audi- does not have a line of sight to the balcony loudspeaker signals are de-ence which is not covered by the bal- loudspeaker cluster . In addi tion, three layed electronical ly so as to synchron-

9


9/11

ize with soun d from the main cen ter I Noise at Receiver Loc. 1 I RACTIV Icluster. Two large "side stack" loud- 500 Hz OB / 2000 Hz OB HAbTi valuespeaker clusters are built-in , one on _. , . . . _ ~~

i „ ,i . . fb,l I b4,/ 0,09each ot the proscenium openings. A

separate multi-cluster loudspeaker 7o' l/5 8'7 0 27syst em is prov ided for coverage of 67,1 /55, 7 0,36tho se sea ted on th e lawn outdoo rs. 64,1 / 52,7 0,42

61,1 /4 9,7 0,46

Th e sound syst ems are used for vir- [ 58,1 / 46,7 | 0,49

tually all popular music presentations. m ,, _ „ A nrT1T , . , , , T , . TOOSOGGBO^ , . ! . \ -. , I able 5. RAbll values in location 1, for different background noiseror classical music, the indoor soundsystems are normally used only for reinforcing instrumental or vocal soloists, but not for ensemble or orchestralreinforcement.

m i . , i . I I RASTIatpos. 15 I RASTIatpos. 16The transmitter was placed stage Forward, Near Center of Main Floor,

center and oriented directly toward Loudspeakers Operating Right LS Stack Right Sidethe audience for acoustical meas ure- ~r „ „ , ~ ~~~ :

, mi i i i m n Center & Balcony Clusters 0,49 0,55ments. the source level used was IhA,+ 10 dB since it is more representative B o t n stacks, Center & 0 5 3 0 52of voice level from a tr ai ne d singer. Balcony ClustersFor measure men ts on the reinfo rce- Both Stacks Alone 0,61 0,58

ment system, a microp hone was ~R ight Stack only 0 6̂8 0 6̂3 placed ab ou t 0,3 m in front of th e 4225 ' ' ' -r„noni ,„n ,loudspeaker. Certain measurements Table 6. Results with different loudspeakerswere made with the sound systemdriven directly by the 4225 electricaloutput.

The receiver's microp hone was the sound system inpu t. Th e purpos e the RAS TI values at different seats placed on a tr ip od so tha t th e micro- of th is pa rt ic ul ar te st was to il lu st ra te serves as a very useful guide in de te r- ph one corr esponded to ea r-he ig ht for th e mask ing effect on inte ll ig ibil ity in mining ar eas of po te nt ia l pr ob le ms .a seated listener. RAS TI meas ure- the hall of rever beran t sound caused Indeed, ther e have been repo rts of in-ment s were made at repr esen tat ive by operati on of art ist s' stage monit or adeq uat e intelligibility during someseat locations thr ough out the main louds peaker s. performan ces in areas with low mea-floor and balcony. No mea sur eme nts sured RAS TI values. Work is unde r

were made outdoo rs. The RAS TI values and backg round way to improve the coverage of thenoise levels as meas ured by the 4419 sound system in thes e areas.Mea sure ment s were made at most receiver at position 1 on the main floor

locations both with and witho ut use of with 3 dB steps in noise from the stage It is also app are nt tha t inapp ropri -the sound amplification system since monito r loudspeake rs are shown in ate signal balance among the variousthe hall may be used in corres ponding Table 5. loudsp eakers can badly dete riora te in-ways. Separate RASTI measu rement s telligibility in certain seating areas.were also mad e with different level RAS TI values were also meas ured For insta nce, opera ting th e balconyadju stme nts of the thre e delayed clus- on the main floor with various combi- loudspe akers at higher levels (+ 6 andters relative to the main cluster. nati ons of loudsp eakers on and off; see + 9 dB) slightly improves the RASTI

Table 6. values at balcony areas 9 and 11, butFig. 6 illus trates the resul ts of the deter iorat es the results on the forward

mea sure ment s. Some care mus t be tak en in inter- part of the main floor - area 1 - as the pr et in g th e re su lt s. Wh en the amplif i- late so und energy from the balcony

The result s indica te tha t intelligibil- cation system is in use the intelligibil- loudspe akers ret urns to the front ofity is rat her low at most seats when no ity depends largely on reverb erati on the hall.amplification is used, as expecte d. charact eristi cs ra the r th an back-Only in the cente r of the room near ground noise. The main building has Ther e are many ways in which athe stage is it possible to und ers tan d reverb erati on charact eristi cs like a complex sound reinforce ment systemunamplifie d speech. fully enclosed hall, but because it is may be opera ted which affect the in-

open to the outdoors , use of uphol- telligibility achieved. The RAS TI

Finally, a set of RAS TI meas ure- stered seats is not practi cal for stabili- metho d allows practi cal objectivements were made at a seat (position 1) zation of rev erbera tion time with analysis of these opera ting metho ds,near the cente r of the main floor when varying audienc e size. This results in as well as compari son with porta blean orches tra was simu late d by playing RAS TI mea sur eme nts in the unoccu- systems which may be brough t in bydifferent levels of rand om noise pied hall which are subst antia lly lower visiting arti sts, or even comparisonthrou gh monit or louds peake rs on tha n would be expected for the more with system s serving other similarstage, but with the 4225 tra nsm itt er usual fully occupied and less reverber- spaces.electrical out put connect ed directly to ant hall. Nevert heles s, compari son of

10

Noise at Receiver Loc. 1RASTI Value

500 Hz OB / 2000 Hz OBRASTI Value

76,1 / 64,7 0,09

73,1 / 61,7 0,19

70,1 /5 8, 7 0,27

67,1 /55,7 0,36

64,1 /5 2, 7 0,42

61,1 /49,7 0,4658,1 /46,7 0,49

T00809GBO

Loudspeakers Operating

RASTI at pos. 15

Forward, Near

Right LS Stack

RASTIatp os. 16

Center of Main Floor,

Right Side

Center & Balcony Clusters 0,49 0,55

Both stacks, Center &

Balcony Clusters0,53 0,52

Both Stacks Alone 0,61 0,58

Right Stack only 0,68 0,63

TonmRdRn


10/11

Conclusions Acknowledgements

Both for the meas urement s de- an objective evaluation of intelligibi l- We want to tha nk all Companiesscribed in thi s application note and ity. and people involved in the measure-for measurements in very large build- ments mentioned above.ings (e.g. St. Paul's Cathedral, see [7]) In future it is expected tha t thethe RASTI method has proven itself RASTI method will become incorpo-to be a very quick and useful method rated in standards for assessing speechfor evaluat ing speech intelligibility. In intelligibi lity in a range of applica-

many circumstances it is the only ex- tions, including the evaluation andisting method which can be used for classification of emergency systems .

References

[1] T. HOUTGAST, H.J.M. STEEN- sion Quality", Jou rnal of Acousti- [6] IEC Publ icat ion 268: Sound Sys-EKEN and R. PLOMP, "Predict- cal Society of America 67, tern Equ ipment Pa rt 16: "Reporting Speech Intelligibility in Rooms 318-326, 1980. on the RASTI method for the ob-

from the Modulation Transfer [4] H.J.M. ST EEN EKE N and T. jective rating of speech intelligibil-Function, I. General Room Acous- HOUTGAST: "RASTI: A Tool for ity in auditoria".

tics", Acustica 46, 60-72, 1980 evaluating auditorial Technical [7] JOH N ANDERSON and TOR-[2] T. HOUTGA ST and H.J .M. Review, No.3, 1985, Bruel & Kjser. BEN JACOBSEN: "RASTI Mea-STEENEKEN, "The Modulation [5] T. HOUTGAST, H.J.M. STE EN- surements in St. Paul 's Cathedra l,Transfer Function in Room EKEN, "A Review of the MTF London" Bruel & Kjser Applica-

Acoustics as a Predictor of Speech Concept in Room Acoustics and tion Note , 1985. Intelligibility", Acustica 28, 66-73, it's use for Estimating Speech In-1973. telligibility in Auditoria' Journal

[3] H.J.M. ST EE NE KE N and T. of Acoustical Society of AmericaHOUTGAST, "A Physical Method 77, 1060-1077, 1985

for Measuring Speech Transmis-

11


11/11

rasti applications

Documents