background image

GOLDBERG AND MÄKIVIRTA

 

OPTIMISED EQUALISATION COMPARISON

 

 

AES 116TH CONVENTION, BERLIN, GERMANY, 2004 MAY 8-11 

16 

 
 
 

Use of Graphic Equaliser - Large models

(incl. 0dB settings)

-12

-11

-10

-9

-8

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

20

25

32

40

50

63

80

10

0

12

5

16

0

20

0

25

0

31

5

40

0

50

0

63

0

80

0

1k

1.

25k

1.

6k

2k

2.

5k

3.

2k

4k

5k

6.

3k

8k

10

k

12

.5

k

16

k

20

k

1/3 Octave Frequency Band, Hz

Level, dB

 

Figure 12a. Use of the graphic equaliser for large systems – including 0dB settings. 
 
 
 

Use of Graphic Equaliser - Large Models

(excl. 0dB settings)

-15

-14

-13

-12

-11

-10

-9

-8

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

20

25

32

40

50

63

80

100

125

160

200

250

315

400

500

630

800

1k

1.

25

k

1.

6k

2k

2.

5k

3.

2k

4k

5k

6.

3k

8k

10k

12.

5

k

16k

20k

1/3 Octave Frequency Band, Hz

Level, dB

 

Figure 12b. Use of the graphic equaliser for large systems – excluding 0dB settings. 
 
 

Summary of Contents for response controls

Page 1: ...t least 40 years an early example is 1 Equalisation is prevalent in professional sound reproduction such as recording studios mixing rooms and sound rein forcement In situ response equalisation is often im plemented using third octave equalisers which are normally set with the help of real time analysers This measurement and equalisation combination is cheap readily available and a relatively simp...

Page 2: ...tudying the statistical properties of 67 in situ magni tude responses before and after equalisation 2 IN SITU EQUALISATION The room response controls were previously described in 12 14 A constant Q type 31 band DSP graphic equaliser 15 was constructed using bi quadratic transfer functions of the form 2 2 1 1 2 2 1 1 0 1 z a z a z b z b b z H 1 where the scaling of the transfer function is given by...

Page 3: ...sponse control settings were calculated for each loudspeaker response Statis tical data was recorded for each magnitude response measurement before and after equalisation to study how the objective quality was improved Further sta tistical analysis is conducted on all measurements in three frequency bands Table 1 LF MF and HF collectively called subbands and correspond ing roughly to the bandwidth...

Page 4: ...adband balance graphic equalisation is able to correct for local fea tures in the response Figure 7 but only with limited success Resonances due to room modes or construc tive interference due to reflections in the response cannot be corrected accurately when the frequencies do not coincide with the centre frequencies of third octave filter bands A good example of this can be seen at 600 Hz In the...

Page 5: ...evels in Subbands Large Systems 6 4 2 0 2 4 6 8 LF MF HF LF MF HF LF MF HF Original Room Reponse Controls Graphic Equaliser Figure 1 Mean and standard deviation of subband median levels before and after room response control and graphic equalisation 5 2 Graphic Equalisation Appendix D Figures 9 13 depicts the use of the equaliser controls for each loudspeaker group The upper graph a shows how the ...

Page 6: ...ow 0 dB indicates that graphic equalisation achieves a response closer to the target For all loud speaker models pooled together the room response controls improved the RMS deviation from 6 1 dB to 4 7 dB improvement 22 whereas graphic equalisa tion improved the RMS deviation to 1 8 dB im provement 70 The main improvement is seen at low frequencies The better performance by the graphic equaliser i...

Page 7: ...ke to thank Louis Fielder for the question in a recent AES Conference in Denmark sparking this paper s content Mr Steve Fisher SCV London for the original inspirational idea for the optimiser and some of the measurements used in the statistical analysis Mr Olli Salmensaari Finnish Broadcasting Corporation for additional measure ments Mr Lars Morset Morset Sound Development and Genelec Oy 9 REFEREN...

Page 8: ...al Commission Geneva 1995 18 The MathWorks MATLAB Optimisation Toolbox v 2 3 The MathWorks Inc Natick 2003 19 Moore B C J Glasberg B R Plack C J and Biswas A K The shape of the Ear s Temporal Win dow J Acoustical Soc America vol 83 pp 1102 1116 1988 Mar 20 Klark Technik http www klarkteknik com 2004 Feb 21 Toole F E Olive S E The Modification of Timbre by Resonances Perception and Measure ment J A...

Page 9: ... 9 APPENDIX A ROOM RESPONSE CONTROL CASE STUDY STATISTICAL GRAPHS Figure 2 Case study optimisation results using room response control equalisation Figure 3 Case study statistical output box plot histogram and normal probability plot before upper and after lower optimised room response control equalisation ...

Page 10: ...Control Equalisation 15 10 5 0 5 Broad LF MF HF RMS Deviation Change due to Room Response Control Equalisation 15 10 5 0 5 Broad LF MF HF Figure 4 Change in sound level deviation due to Room Response Control equalisation for each subband and the broadband quartile difference and RMS of deviation from the broadband median The error bar indicates the stan dard deviation ALL Small Two way Three way L...

Page 11: ...uency Hz Gain dB Centre Frequency Hz Gain dB Centre Frequency Hz Gain dB Centre Frequency Hz Gain dB 20 1 6 200 0 7 2 000 0 0 20 000 0 0 25 6 6 250 0 0 2 500 0 1 32 7 7 315 0 0 3 150 0 0 40 1 4 400 0 0 4 000 0 0 50 2 1 500 0 2 5 000 0 0 63 8 2 630 1 3 6 300 0 0 80 3 3 800 0 0 8 000 0 0 100 0 0 1 000 0 0 10 000 0 0 125 0 0 1 250 0 0 12 500 0 0 160 0 9 1 600 0 0 16 000 0 0 Use of Graphic Equaliser 1...

Page 12: ...6TH CONVENTION BERLIN GERMANY 2004 MAY 8 11 12 Figure 7 Case study optimisation results using graphical equalisation Figure 8 Case study statistical output box plot histogram and normal probability plot before upper and after lower optimised graphical equalisation ...

Page 13: ...k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz Level dB Figure 9a Use of the graphic equaliser for small 2 way systems including 0dB settings Use of Graphic Equaliser Small models excl 0dB settings 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 2 3 20 25 32 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1k 1 25k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 ...

Page 14: ... 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz Level dB Figure 10a Use of the graphic equaliser for 2 way systems including 0dB settings Use of Graphic Equaliser 2 way models excl 0dB settings 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 2 3 20 25 32 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1k 1 25k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz...

Page 15: ... 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz Level dB Figure 11a Use of the graphic equaliser for 3 way systems including 0dB settings Use of Graphic Equaliser 3 way models excl 0dB settings 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 2 3 20 25 32 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1k 1 25k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz...

Page 16: ... 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz Level dB Figure 12a Use of the graphic equaliser for large systems including 0dB settings Use of Graphic Equaliser Large Models excl 0dB settings 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 2 3 20 25 32 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1k 1 25k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz...

Page 17: ...2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz Level dB Figure 13a Use of the graphic equaliser for all systems including 0dB settings Use of Graphic Equaliser All Models excl 0dB settings 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1 2 3 20 25 32 40 50 63 80 100 125 160 200 250 315 400 500 630 800 1k 1 25k 1 6k 2k 2 5k 3 2k 4k 5k 6 3k 8k 10k 12 5k 16k 20k 1 3 Octave Frequency Band Hz L...

Page 18: ...Graphic Equalisation 15 10 5 0 5 Broad LF MF HF RMS Deviation Change due to Graphic Equalisation 15 10 5 0 5 Broad LF MF HF Figure 14 Change in sound level deviation due to graphic equalisation for each subband and the broadband quartile difference and RMS of deviation from the broadband median The error bar indicates the standard deviation ALL Small Two way Three way Large ALL Small Two way Three...

Page 19: ...on Change between Equalisations 15 10 5 0 5 Broad LF MF HF Figure 15 The difference between the change in sound level deviation for the room response control and the graphic equalisation techniques for each subband and the broadband quartile difference and RMS of deviation from the broadband median are plotted A value below 0dB indicates that graphic equalisation achieves a response closer to the ...

Reviews: