The Effect of the Acoustics of Sound Control Rooms on the Perceived Acoustics of a Live Concert Hall Recording

Transcription

1 The Effect of the Acoustics of Sound Control Rooms on the Perceived Acoustics of a Live Concert Hall Recording C.C.J.M. HAK 1 Architecture, Building and Planning, unit BPS Laboratorium voor Akoestiek, Eindhoven University of Technology Den Dolech, 5600 MB Eindhoven The Royal Conservatoire, deartment Art of Sound Juliana van Stolberglaan 1, 595 CA The Hague THE NETHERLANDS c.c.j.m.hak@tue.nl Abstract: - Live recordings of music and seech in concert halls have acoustical roerties, such as reverberation, definition, clarity and saciousness. Sound engineers lay back these recordings through loudseakers in sound control rooms for audio CD or film. The acoustical roerties of these rooms influence the erceived acoustics of the live recording. To find the ractical imact of a sound control room on the acoustical arameter values of a concert hall, combinations of concert hall imulse resonses and sound control room imulse resonses have been investigated using convolution techniques. It can be concluded that the ITU-recommendations used for sound control room design are sufficient for reverberation and seech intelligibility judgement of concert hall recordings. Clarity judgement needs a very high decay rate, while judgement of saciousness can only be done by headhone. Key-Words: Sound control, Sound studio, Control room, Room acoustics, Concert hall, Recording, Playback, Convolution, Head and torso simulator, HATS 1 Introduction From exerience and an earlier investigation [1] it is clear that a recorded reverberation time can only be heard in a room having a reverberation time shorter than the one in which the recording was made. The smallest details and the finest nuances with regard to colouring, definition and stereo image can only be judged and criticized when there is little acoustical influence from the layback acoustics on the recorded acoustics. However, usually the listening or layback room in combination with the used sound system affects the recorded acoustics. This haens in class rooms, congress halls, cinemas and even in sound control rooms. The imact of the control room acoustics on live recorded acoustics has been investigated, using the acoustic measurement rogram DIRAC. To this end the convolution function has been alied to binaural imulse resonses of six control rooms, a symhonic concert hall, a chamber music hall and a rofessional headhone. The imact on reverberation, seech intelligibility, clarity and inter-aural cross-correlation has been investigated. From the results, a first ste is made to judge the quality of a sound control room using this new aroach, starting from the JND (Just Noticeable Difference) as allowable error. Procedure Starting from a set of 6 binaural control room imulse resonses, 1 binaural headhone imulse resonse and 4 binaural concert hall imulse resonses, 8 airs of imulse resonses are defined. From each air (h 1, h ) the first is considered as a concert hall imulse resonse and the other as a control room imulse resonse. Each air (h 1, h ) is convolved (see Section 4) to obtain the imulse resonse h 1, heard when laying back the recorded concert hall imulse resonse in the sound control room. h 1, thus reresenting h 1 affected by h, is then comared to h 1, with resect to the reverberation time T 30, the clarity C 80, the modulation transfer index MTI and the inter-aural cross-correlation coefficient IACC [,3]. 3 Room acoustic arameters Many room acoustic arameters are derived from the room s imulse resonses. Examles of such arameters are the reverberation time, which is related to the energy decay rate, the clarity, the definition and the centre time, which are related to early to late energy ratios, the seech intelligibility, which is related to the energy modulation transfer characteristics of the imulse resonse and the latereral energy fraction, the late lateral sound energy and the inter-aural cross-correlation, which are related to the lateral imulse resonse measurements. ISSN: ISBN:

2 Four of them have been investigated, being the reverberation time T 30, the clarity C 80, the modulation transfer index MTI and the inter-aural cross-correlation. The JND-values of these arameters are resented in table Reverberation time T The reverberation time T is calculated from the squared imulse resonse by backwards integration [4] through the following relation: [ db] t L( t) = 10lg (1) 0 where L(t) is the equivalent of the logarithmic decay of the squared ressure. For this investigation the T 30 with its evaluation decay range from -5 db to -35 db is used to determine T. 3. Clarity C 80 The arameter C 80 [5] is an early to late arriving sound energy ratio intended to relate to conditions for music and is calculated from the imulse resonse using the following relation: C 80ms 0 80ms [ db] = 10 () 80 lg 3.3 Modulation Transfer Index The Modulation Transfer Function m(f) [6] describes to what extent the modulation m is transferred from source to receiver, as a function of the modulation frequency F, which ranges from 0.63 to 1.5 Hz. The m(f) is calculated from the squared imulse resonse using the folowing relation: m( F ) = e jπft [ ] (3) The m(f) values for 14 modulation frequencies are averaged, resulting in the so called modulation transmission index MTI [7], given by: 14 m( F n ) n= 1 MTI ( F ) = 14 [ ] (4) 3.4 Inter-aural cross-correlation coefficient IACC Although the IACC is still subject to discussion and research, the arameter IACC [8] is used to measure the satial imression and is calculated from the imulse resonse using the following relation: IACF t ( t + τ ) l r t t1, t t t l r t1 t 1 ( τ ) = [ ] (5) where 1 (t) is the imulse resonse measured at the left ear and r (t) is the imulse resonse measured at the right ear of the HATS. The inter-aural cross-correlation coefficient IACC is given by: IACCt 1,t = IACFt 1,t (τ) max for -1ms < τ <+1ms (6) For this investigation only the interval between t 1 = 0 and t = 80 ms (early reflections) is used. Table 1. JND (Just Noticeable Difference). T 30 C 80 MTI IACC 10 % 1 db Imulse resonses and measurements 4.1 Measurement conditions All measurements, both the single channel and the dual channel, were erformed using a HATS or an artificial head [9]. The decay range (INR) [10] for all measured imulse resonses had a minimum exceeding 5 db for all octave bands used Large and small concert hall Imulse resonse measurements were erformed in the large and small concert hall of The Frits Philis Muziekcentrum Eindhoven with a volume of arox. 14,400 m 3, an unoccuied stage floor and T emty s for the large hall and a volume of arox m 3, an unoccuied stage floor and T emty 1.5 s for the small (chamber music) hall. Figures 1 en give an imression of the halls and the schematic floorlans with the source osition S as indicated, laced on the major axis of the hall, and the microhone ositions R1 and R, where R1 is laced at arox. 5 m from the source, equal to the critical distance, and R is laced at arox. 18 m (diffuse field). More secifications of both concert halls are resented in table, using the total average over both microhones (ears) of the HATS, the 500 and 1000 Hz octave bands and the receiver ositions R1 and R. The INR for all measured symhonic and chamber music hall imulse resonses had an average of 60 db for all used octave bands, with a minimum exceeding 54 db. ISSN: ISBN:

3 Fig 1. Measured symhonic and chamber music hall. Fig 3. Concert hall measurement using a Head and Torso Simulator (HATS). Table. Concert hall secifications Control rooms The control rooms under test are all Dutch control rooms and qualified as very good by the sound engineers as well as the designers. For the sake of rivacy the measured control rooms are marked from CR1 to CR6. The control rooms were measured extensively with microhone ositions laced on a grid consisting of 15 measurement ositions [11]. Based on these measurements several imortant room acoustical arameters were comuted. The results of the reverberation time measurements revealed that in the lower frequencies all control rooms under test met the generally used ITU recommendation [1]. In the higher frequencies only control room CR met this criterion. Secifications of all control rooms under test are resented in table 3. The control room imulse resonse measurements were erformed at the sound engineer osition, known as the sweet sot, the focal oint between the main (wall mounted) loudseakers. Control room CR5 was only suitable for near field monitoring. The INR for all measured control room imulse resonses had an average of 57 db for all used octave bands, with a minimum exceeding 5 db. Fig. Sound source S and microhone R ositions. Fig 4. Sweet sot measurement using a Head and Torso Simulator (HATS) or an artificial head. ISSN: ISBN:

4 Table 3. Control room secifications. Head And Torso Simulator: used in concert halls and control rooms. (B&K - Tye 418C); artificial head: used in control rooms (Sennheiser - MZK 00) microhones: used with artificial head (Sennheiser - MKE 00) ower amlifier: used in concert halls (Acoustics Engineering - Amhion); sound source: omnidirectional, used in concert halls. (B&K - Tye 49); sound device: USB audio device (Acoustics Engineering - Triton); headhone: used as a reference source; (Philis: SBC HP890) measurement software: DIRAC (B&K - Tye 7841) 5 Convolution The convolution y of signal s and system imulse resonse h is written and defined as: or y s h( t) = (7) y ( s h) = s h( t τ ) dτ = (8) 4.1. Headhone To comlete the set of imulse resonses a ure free field measurement was erformed using the HATS and a high quality headhone as shown in figure 5. The minimum INR for the measured imulse resonse reached a value of 88 db for both the 500 Hz and the 1 khz octave band. Fig 5. Headhone transfer measurement using a HATS. 4. Measurement equiment The measurement equiment consisted of the following comonents: In words: the convolution is defined as the integral of the roduct of two functions s and h after one is reversed and shifted. From a room acoustical oint of view s(t) is a iece of music that is recorded in a concert hall and layed back in a sound control room, h(t) the imulse resonse of the control room and y(t) the convolved sound as it is heard in that control room. When the control (listening) room is reverberant, smoothing of the sound occurs. The room acoustics in the music recording that we want to judge will be affected by the acoustics of the sound control (listening) room. With a double convolution by which an imulse resonse from the concert hall is convolved with a dry recording of music and afterward the result is convolved with the imulse resonse of a sound control room, it is ossible to hear how a recording, made in the concert hall, sounds when layed back in the control room. The result is a more or less smoothed sound signal. By using a ure imulse (Dirac delta function) instead of a normal sound signal to be convolved with both room imulse resonses (eq 9 and 10) we can examine what the control room does with the concert hall concerning the values for the room acoustic arameters (eq 11). Therefore it is ossible to derive a room in room acoustic arameter value from the more or less smoothed imulse resonse. Mathematically: ISSN: ISBN:

5 = δ h h h δ = (9) Where: h(t) = room imulse resonse δ(t) = Dirac delta function (ideal imulse) h h = h h h1 1 1 = δ (10) Where: h 1 (t) = total imulse resonse room1 room h 1 (t) = imulse resonse room 1 h (t) = imulse resonse room Substituting equation (10) into equation (7) results in: s h y1 1 = (11) Where: y 1 (t) = convolution of a random sound signal with the total imulse resonse s(t) = random sound signal 6 Measurement results In Figure 6 through 13 the results of the convolutions are deicted as an average over the 500 and 1000 Hz octave band. Each grah shows the difference between values of a arameter, one calculated from h 1, the convolution of the concert hall with the control room and one from h 1, the imulse resonse of the concert hall. On the x-as the control rooms CR1 to CR6 are given in order of the decay rate. The differences are calculated for four acoustical arameters: T 30, C 80, MTI and IACC. Fig 8. Difference between C 80h1 and C 80h1 (C 80 error) measured at (hall) osition R1.. Fig 9. Difference between C 80h1 and C 80h1 (C 80 error) measured at (hall) osition R. Fig 10. Difference between MTI h1 and MTI h1 (MTI error) measured at (hall) osition R1. Fig 6. Percentual difference between T h1 and T h1 (T 30 error) measured at (hall) osition R1. Fig 11. Difference between MTI h1 and MTI h1 (MTI error) measured at (hall) osition R. Fig 7. Percentual difference between T h1 and T h1 (T 30 error) measured at (hall) osition R. ISSN: ISBN:

6 Fig 1. Difference between IACC h1 and IACC h1 (IACC error) measured at (hall) osition R1. Fig 13. Difference between IACC h1 and IACC h1 (IACC error) measured at (hall) osition R. 7 Conclusion Starting with six qualified good, more or less standardised sound control rooms, two concert halls and the Just Noticeable Difference of four calculated room acoustic (ISO/IEC) arameters it can be concluded: The ITU-recommendations for sound control room design are adequate for evaluation of reverberation in concert hall recordings. When it is imortant to asses the details of sound definition of a concert hall recording, you need a control room with a very high decay rate. Only the control rooms with a reverberation time below 0.15 s can be used. The ITU-recommendations for sound control room design are adequate for evaluation of seech intelligibility in concert hall recordings. An accurate judgement of saciousness of a binaural concert hall recording aarently requires the use of headhones. This requires further investigation. ISO/DIS Draft: 006 (International Organization for Standardization, Geneva, Switzerland, 006). [3] Sound System Equiment Part 16: Objective rating of seech intelligibility by seech transmission index, International Standard IEC : 003 (International Organization for Standardization, Geneva, Switzerland, 003). [4] M.R. Schroeder, Integrated-imulse method for measuring sound decay without using imulses, Journal of the Acoustical Society of America, 66, (1979) [5] W.Reichar, O. Abdel Alim, W. Schmi, Definition und Meβgrundlage eines objektiven Maβes zur Ermittlung der Grenze zwischen brauchbarer und unbrauchbarer Durchsichtigkeit bei Musikdarbitung, Acustica 3, (1975). [6] M.R. Schroeder, Modulation tranfer functions: definition and measurements, Acustica, 49, (1981). [7] T. Houtgast, H.J.M. Steeneken, The modulation transfer function in room acoustics as a redictor of seech intelligibility, Acustica, 8, (1973). [8] W.V. Keet, The influence of early reflections on satial imressions, 6 th ICA Tokyo Jaan (1968). [9] C.C.J.M. Hak, M.A.J. v. Haaren, R.H.C. Wenmaekers, L.C.J. van Luxemburg, Measuring room acoustic arameters using a head and torso simulator instead of an omnidirectional microhone, Internoise 009, Ottawa, Canada (009). [10] C.C.J.M. Hak, J.P.M. Hak, R.H.C. Wenmaekers, INR as an Estimator for the Decay Range of Room Acoustic Imulse Resonses, AES Convention Amsterdam (008). [11] B.J.P.M. v. Munster, Beyond Control Acoustics of Sound Recording Control Rooms Past, Present and Future, FAGO Reort nr: 03-3-G, Unit: Building Physics and Systems, Deartment of Architecture, Building and Planning, University of Technology Eindhoven (003). [1] ITU-recommendation, Methods for the Subjective Assessment of Small Imairments in Audio Systems including Multichannel Sound Systems, ITU-R BS (1997). References: [1] C.C.J.M. Hak, R.H.C. Wenmaekers, The effect of Room Acoustics on the Perceived Acoustics of Reroduced Sound, Internoise 008, Shanghai, China (008). [] Acoustics-Measurement of room acoustic arameters - Part 1: Performance rooms, International Standard ISSN: ISBN: