Sunteți pe pagina 1din 12

 

 
Pinecone Research Labs 
How  timbral  quality  of  sound  is  affected  
when  using  binaural  worldizing  
techniques.  

Benjamin Costerton 

 
 
 
January, 2018 

 
 

How  timbral  quality  of  sound  is  affected   example  mono  instrument  recordings  were  sent  
when  using  binaural  worldizing   to  a  large,  mostly  empty,  room  somewhere  
within  the  studio  and  played  out  of  a  
techniques.     loudspeaker.  Also  in  the  room  would  be  a  mono  
  microphone  recording  the  instrument  back,  
  along  with  all  the  characteristics  of  the  large  
Abstract   room.    
   
The  research  aims  to  establish  to  what  extent   The  worldizing  technique  continued  to  be  
timbral  quality  is  affected  when  a  binaural   popular  during  the  production  of  Star  Wars  in  the  
worldizing  technique  is  used  to  manipulate  the   late  1970's  as  Ben  Burtt  has  spoken  about  his  
spatial  properties  of  audio.     use  of  the  technique  for  adding  a  more  
  "authentic"  sound  to  elements  of  the  soundtrack.  
Two  experiments  on  different  source  material   "I  remember  many  hours  playing  back  Darth  
are  carried  out  both  in  order  to  answer  the   Vader's'  voice  in  offices,  hallways,  and  
above  question,  but  also  gain  some  insight  into   bathrooms,  in  order  to  get  just  the  right  acoustic  
the  most  appropriate  equipment  and   quality"  ​[3]​.    
configurations  for  binaural  worldizing.     Erik  Aadahl  and  Ethan  Van  der  Ryn  also  
  commented  on  their  use  of  the  worlizing  
The  research  found  that  in  some  cases  spatial   technique  during  the  production  of  Godzilla  
awareness  and  a  sense  of  externalisation  can   (2014)  [4].  ​"We  actually  ended  up  renting  the  
be  achieved  using  a  binaural  worldizing   tour  speaker  array  for  the  Rolling  Stones,  which  
technique,  however  in  other  cases  there  were   was  about  10-­feet  tall  and  about  18-­feet  
negative  effects  to  the  resulting  audio.     wide...we  played  all  the  creature  vocals  through  
  that  and  then  recorded  it  through  cars,  from  the  
The  research  presents  some  guidelines  for   other  side  of  the  lot,  from  rooftops,  from  within  
applying  a  binaural  re-­recording/worldizing   storefronts,  so  we  could  get  that  real-­world  
technique  and  some  recommendations  on   feeling  of  what  it  would  be  like  to  be  on  a  city  
equipment.     street  and  have  Godzilla  50  feet  away".​This  
  example  is  on  a  much  larger  scale,  but  shows  
           ​  1.  What  is  binaural  worlizing?     how  the  technique  of  worldizing  was  adapted  to  
  achieve  the  needed  results,  in  this  case  getting  
Worlizing  is  a  recording  technique  first  coined  by   a  speaker  array  as  big  as  Godzilla.    
Walter  Murch  back  in  1973  while  working  on    
American  Graffiti  (1973)  [1].  The  technique   Binaural  worldizing  is  another  adaptation  on  the  
involves  playing  clean,  generally  unprocessed   worldizing  technique,  whereby  sound  is  played  
sound,  through  a  speaker  in  a  real  world   from  a  speaker  into  the  real  world,  and  recorded  
location,  and  recording  it  back  with  some  sort  of   back  using  binaural  recording  techniques.  The  
portable  recording  setup.  The  reason  for  doing   main  purpose  of  this  is  to  capture  sound  in  a  
so  it  that  you  now  have  a  recording  of  the   binaural  format  so  that  it  can  be  listened  back  to  
original  source  sound,  but  with  all  the  location   using  headphones  to  achieve  a  3D  effect.    
cues  of  the  space  you  recorded  in.     The  main  focus  is  on  achieving  the  3D  effect  of  
You  could  argue  that  the  technique  dates  further   binaural,  but  also  the  room  the  sound  is  
back  to  the  use  of  echo  chambers  in  early  studio   recorded  in  will  affect  the  timbral  quality  of  the  
recordings.  One  famous  example  being  the  use   sound.  This  paper  aims  to  see  to  what  extent  the  
of  an  echo  chamber  on  the  1959  recording  of   timbral  quality  of  sound  is  affected  once  a  
Miles  Davis'​  ​Kind  of  Blue​  ​(1959)  recorded  at   binaural  worlizing  technique  is  used.    
Columbia  records  30th  street  studio  [2].  In  this    
 

1
 

  microphones  all  audio  being  recorded  is  


         ​  2.  What  is  timbral  quality?   subjected  to  one  set  of  HRTFʼs  and  the  outcome  
  may  not  translate  well  to  every  listener.    
When  the  research  refers  to  timbre,  it  is  referring    
to  the  perceived  quality  of  the  sound.  In  music  it              ​4.  Known  issues  
is  sometimes  refered  to  as  the  colour  or  tone  of    
the  music.  The  research  will  refer  to  several   A  binaural  re-­recording,  or  worldizing,  could  
elements  that  determine  the  timbral  quality  of   create  a  3D  image,  and  depending  on  the  room  
the  sound  effects  in  this  paper,  they  are;;   could  capture  a  good  sense  of  externalisation,  
  but  unfortunately  the  same  issues  surrounding  a  
-­ Frequency   one-­size-­fits-­all  HRTF  recording  apply  with  
-­ Dynamics   worldizing.    
-­ Room  Reflections   Many  end-­users  will  undoubtedly  experience  
  issues  because​  "it's  not  your  own  pinna,  and  you  
Frequency:   are  having  to  use  someone  else's  ears  
Changes  to  any  frequency  band  that  is  affecting   effectively"  ​[6]​.  ​Using  someone  else's  pinna  can  
the  tone  of  the  sound.   cause  confusion  and  timbral  issues  for  many  
  user.  As  the  model  used  in  this  experiment  uses  
Dynamics:   a  single  mould  of  human  pinna,  it  will  not  be  an  
Whether  the  amplitude  or  dynamic  range  is   exact  match  for  every  listener.    
negatively  affected.  Negative  timbral  quality    
could  be,  in  part,  a  result  of  a  reduction  in    
dynamic  range  of  a  sound.            ​  5.  Stereo  rainfall  example  
   
Reflections:   The  attached  rainfall  example  consists  of  three  
Not  exactly  a  quality  of  the  source  material,  but   recordings  made  using  the  binaural  worldizing  
binaural  worldizing  will  add  room  reflections,  and   technique.  The  microphones  used  for  the  
this  can  add  space  to  a  recording.  Space  is  a   binaural  recording  consists  of  a  pair  of  LUHD  
desired  characteristic  in  worldizing,  but  it  also   PM-­01AB  [7]  omnidirectional  microphones  
affects  the  timbral  quality.     placed  either  side  of  a  model  head  featuring  
  silicone  moulds  of  human  pinna.    
 
           ​3.  Existing  examples  
 
In  2010  Jorge  Miramontes,  a  researcher  from  
Stanford  University  recorded  the  output  from  a  
5.1  speaker  setup  using  in-­ear  binaural  
microphones  [5].  The  microphones  used  were  
Sound  Professionals  SP-­TFB-­2,  an  in-­ear  stereo  
microphone  setup  used  to  simulate  the  binaural  
effect.  The  recording  was  made  of  the  entire  5.1  
mix  as  one  single  recording  in  a  studio  
environment.  
Whether  the  two  excerpts  of  film  recorded  in  this  
fashion  are  able  to  achieve  a  sense  of  space  or  
not  would  have  to  be  decided  through  trials  to  
establish  an  average  opinion.  As  the  research  
will  continue  to  explain,  when  using  binaural  
 

2
 

approximately  10cm  from  the  ear  on  both  sides  


of  the  head  model.    
 
Rainfall  in  binaural  far.wav  ​-­  A  binaural  
worldizing  recording  made  using  a  pair  of  
Harbeth  M20  speakers  spaced  30cm  apart.  The  
setup  in  this  examples  replicates  that  of  the  
'​Rainfall  in  binaural  close.wav'​  example  with  the  
exception  of  the  speakers  being  30cm  from  the  
head  models  'ears'.    
   
 
           ​5.1  Spectral  audio  analysis    
 

 
Binaural  microphone  whilst  recording  ​'Rainfall  in  binaural  
close.wav'  
 
The  files  in  this  example  are  as  follows;;    
  Rainfall  NOT  binaural.wav  
Rainfall  NOT  BINAURAL.wav​  -­  A  stereo  
recording  made  using  an  XY  stereo  
microphones.  ​This  is  the  recording  played  out  
of  the  speakers  in  the  binaural  worldizing  
test.  
 
Rainfall  in  binaural  mono.wav  ​-­  A  binaural  
worldizing  recording  made  using  a  mono  
speaker  playing  the  rainfall  sfx  at  0​°  azimuth,    
45​°  elevation,  and  10cm  from  the  microphone.   Rainfall  in  binaural  mono.wav  
The  speaker  used  in  this  test  was  a  Yamaha  
MS101ii  [8].    
 
Rainfall  in  binaural  close.wav  ​-­  A  binaural  
worldizing  recording  made  using  a  pair  of  
Harbeth  M20  [9]  speakers  spaced  30cm  apart.  
The  speaker  set-­up  is  designed  to  create  a  
phantom  image  at  0​°  azimuth  and  45°  elevation.  
The  elevation  effect  was  created  by  moving  the  
Rainfall  in  binaural  close.wav  
microphone  into  a  45°  elevated  angle  position  
instead  of  moving  the  speakers  as  this  is  easier  
given  the  setup.  See  image  above.  This  
recording  was  made  with  each  monitor  being  

3
 

100-­400​  Hz,  with  only  a  small  lack  in  frequency  


response  below  50  Hz.    
   
           ​5.1.1  Changes  to  dynamic  range  
 
The  dynamic  range  of  the  recordings  were  
tested  using  a  TC  Electronics  LM6n.  The  
dynamic  range  of  the  recordings  are  as  follows;;  
   
Rainfall  in  binaural  far.wav     Rainfall  NOT  binaural.wav  -­  10.9  dB  
  Rainfall  in  binaural  mono.wav  -­  3.7  dB  
'Rainfall  in  binaural  mono.wav'  shows  some   Rainfall  in  binaural  close.wav  -­  15.3  dB  
clear  negative  effects  on  frequency  response   Rainfall  in  binaural  far.wav  -­  8dB  
compared  to  the  original  recording.  The  speaker    
used,  a  Yamaha  MS101ii,  was  not  able  to   'Rainfall  NOT  binaural.wav'  being  the  original  
reproduce  the  low  end  frequencies  in  the   recording,  and  all  subsequent  recordings  being  
recording  at  around  100-­300  Hz,  and  there  is   binaural  worldizing  recordings.    
also  a  clear  drop  off  of  response  above  10  kHz.      
This  is  shown  clearly  in  the  spectral   The  only  correlation  that  can  be  made  between  
representation  of  the  sound  as  well  as  it  being   the  different  examples  presented  here  is  a  
clearly  audible.     reduction  in  dynamic  range  as  the  microphone  is  
  moved  further  from  the  speakers.    
 
As  '​Rainfall  in  binaural  far.wav'  ​has  a  dynamic  
range  of  8dB,  it  has  a  reduced  dynamic  range  
compared  to  the  original  non-­binaural  recording  
at  10.9dB.  '​Rainfall  in  binaural  close.wav'  ​has  a  
greater  dynamic  range  when  compared  to  the  
non-­binaural  recording  at  15.3dB.  
The  above  suggests  that  placing  the  microphone  
somewhere  between  the  '​close'​  position  of  10  
cm  and  the  '​far'  ​position  of  30  cm  would  achieve  
a  matched  dynamic  range  to  the  original  
  non-­binaural  recording.    
Binaural  microphone  tilted  to  simulate  Harbeth  M20      
speakers  position  above  the  microphone  at  a  45​°  elevation    
             ​6.  Worldizing  moving  sound  cues  
Both  'Rainfall  in  binaural  close.wav'  and  'Rainfall    
in  binaural  far.wav'  were  recorded  using  Harbeth   A  second  test  was  performed  to  see  how  moving  
M20  speakers  in  the  set  up.  They  are  also  both   'one-­shot'  style  sound  cues  would  translate  to  
played  back  in  a  2.0  stereo  configuration.  Both   binaural  worldizing.    
of  these  recordings  show  good  frequency    
response  up  to  15  kHz,  after  which  there  is  a   The  SFX  used  in  the  test  were  two  versions  of  a  
sharp  drop  off  in  response.  Both  recordings  also   'woosh'  SFX,  often  used  as  an  accent  to  an  
show  a  very  acute  lack  in  response  at  1.5  kHz.   action  or  movement  when  used  as  part  of  a  
The  thunder  in  the  recording  is  very  clear  and   soundtrack.  The  sound  effect  was  chosen  as  a  
audible  in  both  recordings  at  around   contrast  to  the  atmospheric  soundscape  
provided  by  the  rainfall  test.    
 

4
 

  triangle  with  the  dummy  head  in  a  forward  facing  


position.    
 
Short  woosh  L  to  R.wav  
Same  setup  as  above,  but  using  the  shorter  
woosh  sfx.    
 
Long  woosh  Front  to  Rear.wav  
Harbeth  speakers  were  placed  directly  behind  
and  infront  of  the  dummy  head.  
 
Short  woosh  Front  to  Rear.wav  
Same  setup  as  above,  but  using  the  shorter  
 
Binaural  microphone  recording  sound  cues  from  a  Harbeth   woosh  sfx.    
M20  in  the  RR  position  of  a  quadraphonic  setup.    
   
   
The  files  in  this  example  are  as  follows;;              ​6.1  Spectral  audio  analysis  
   
Long  woosh  pass-­by.wav  
A  stereo  sound  effect  of  a  'woosh'  panning  
slowly  from  right  (R)  to  left  (L).  ​This  is  one  of  
two  recordings  played  out  of  the  speakers  in  
the  binaural  worldizing  test.  
 
Short  woosh  pass-­by.wav  
A  shorter  stereo  sound  effect  of  a  'woosh'  
 
panning  quickly  from  right  (R)  to  left  (L).  ​This  is  
Long  woosh  pass-­by.wav  
one  of  two  recordings  played  out  of  the  
speakers  in  the  binaural  worldizing  test.  
 
 
Long  woosh  RR  to  RF.wav  
Using  a  quadraphonic  speaker  arrangement,  
Harbeth  M20  speakers  were  placed  at  the  
right-­rear  (RR)  and  right-­front  (RF)  positions,  
forming  an  equilateral  triangle  with  the  dummy    
heads  right  ear.     Short  woosh  pass-­by.wav  
 
Short  woosh  RR  to  RF.wav  
Same  setup  as  above,  but  using  the  shorter  
woosh  sfx.    
 
Long  woosh  L  to  R.wav  
Using  a  quadraphonic  speaker  arrangement,  
Harbeth  M20  speakers  were  placed  at  the  left  
(L)  and  right  (R)  positions,  forming  an  equilateral  

5
 

   
Long  woosh  RR  to  RF.wav   Long  woosh  Front  to  Rear.wav  

   
Short  woosh  RR  to  RF.wav   Short  woosh  Front  to  Rear.wav  
 
Across  all  of  the  binaural  recordings  there  are  
clear  differences  in  frequency  response  which  is  
to  be  expected  when  adapting  sound  to  a  
binaural  configuration,  and  similar  results  to  the  
previous  test  in  section  5  of  this  paper.    
There  are  reductions  in  the  overall  amplitude  of  
the  binaural  recordings  when  compared  to  the  
  original,  again  these  are  similar  results  to  what  
Long  woosh  L  to  R.wav   was  seen  in  the  previous  test  in  section  5.    
 
6.1.1  Frequency  
 
In  all  the  binaural  tracks  there  is  a  clear  drop  off  
of  frequency  response  at  around  10-­12kHz.  The  
Harbeth  M20's  specify  a  75Hz  to  20kHz  
frequency  range,  but  unfortunately  LUHD  do  not  
quote  a  frequency  response  of  the  ​PM-­01AB.  
 
Therefore  the  research  is  unable  to  state  
Short  woosh  L  to  R.wav  
whether  this  loss  in  upper  frequency  response  is  
due  to  the  recording  technique  or  the  frequency  
response  on  the  microphones  used.    
 
Important  frequencies  for  sound  localisation  
however  are  present  in  all  binaural  worldizing  
examples  given.  The  human  voice,  being  an  
important  part  of  a  film  mix  or  song  features  
across  the  frequency  spectrum  but  is  most  

6
 

intelligible  at  around  1-­3kHz​  ​[10],  and  these   such  as  ​'Short  woosh  L  to  R.wav'​  feature  audible  
important  frequencies  remain  intact  in  all   additions  of  room  tone  that  change  the  spatial  
binaural  worldizing  examples.   awareness  of  the  recording.  
Frenchies  below  700Hz,  the  point  where  the    
distance  between  an  average  pair  of  ears  is   6.1.4  Panning  
equal  to  half  the  wavelength,  play  an  important    
role  in  the  localisation  of  sound  cues  [11].  As  the   Some  binaural  examples  such  as  '​Short  woosh  L  
frequency  spectrum  images  show,  there  is  good   to  R.wav​'  feature  good  levels  of  spatial  
frequencies  response  in  all  binaural  audio   positioning  with  clear  movement  of  the  sound  
between  0Hz-­700kHz.     cues  from  Left  to  Right.  Binaural  recordings  
As  these  two  important  spectrums  of   featuring  sound  cues  panning  from  a  front  
frequencies  are  present  after  the  binaural   position  to  a  rear  position  are  not  as  easy  to  
worlizing  has  taken  place,  localisation  and   position  in  space.  This  is  most  likely  down  to  the  
intelligibility  of  sound  cues  remains  intact.     known  issues  surrounding  front/back  
  localisation.    
6.1.2  Dynamics    
  6.1.5  Front/Back  Confusion  
Following  from  the  results  of  the  rainfall  tests  in    
section  5  of  this  paper,  the  dummy  head  was   Binaural  recordings  that  feature  sound  cues  
placed  at  20cm  from  the  Harbeths  for  all  test,  in   moving  from  a  position  behind  the  listener  to  the  
an  attempt  to  match  the  dynamic  range  of  the   front  of  the  listener  suffer  from  front/back  
original  recordings.     confusion.    
  Both  'Front  to  Rear'  and  'RR  to  FR'  suffer  from  
The  dynamic  range  of  the  woosh  recordings   front/back  confusion  and  features  less  obvious  
were  tested  using  a  TC  Electronics  LM6n.  The   localisation  cues.    
dynamic  range  of  the  recordings  are  as  follows;;   As  the  recordings  offer  no  opportunity  for  the  
  listener  to  use  head  movement  to  assist  in  
Long  woosh  pass-­by.wav  -­    8.9dB   localisation,  front/back  confusion  is  expected  in  
Short  woosh  pass-­by.wav  -­  10.3dB   the  recordings  [11,12]    
Long  woosh  RR  to  RF.wav  -­  8.1dB    
Short  woosh  RR  to  RF.wav  -­  12.6dB    
Long  woosh  L  to  R.wav  -­  10.8dB   7  Results  
Short  woosh  L  to  R.wav  -­  12.7dB    
Long  woosh  Front  to  Rear.wav  -­  10.7dB    
Short  woosh  Front  to  Rear.wav  -­  11.9dB      
   
Using  the  20cm  distance  from  speakers  to    
microphone  array  has  kept  changes  in  dynamic    
range  to  around  2dB  in  all  cases,  an    
improvement  on  the  rainfall  test  in  section  5  of    
this  paper.      
   
6.1.3​  ​Spatial  Awareness  /  Room  Reflections    
   
The  spectral  analysis  of  the  binaural  recordings    
show  the  addition  of  room  tone  in  all  cases.  This    
room  tone  however  is  not  at  a  level  where  it  is    
negatively  affecting  the  recordings.  Recordings    
 

7
Fig  1.  Table  outlining  the  results  of  the  rainfall  binaural  worldizing  tests.    
 
File   Pros   Cons   Comments  

Rainfall  NOT   Good  spatial  (left/right)   Lack  of  externalisation,   Original  non  binaural  
BINAURAL.wav   separation  due  to  XY   probably  due  to  nature   audio  used  for  
microphone  pattern   of  recording,  being  a   playback  in  
used.   large  ambient   worldizing  
soundscape.     experiment.    

Rainfall  in  binaural     Bad  representation  of   This  recording  tests  a  


mono.wav   original  sound.  A  lot  of   different  speaker  aimed  
frequency  colouration  -­   at  highlighting  the  
Lack  of  low-­end.   difference  in  timbre  
Greatly  reduced   achieved  by  using  
dynamic  range  from   different  speakers.  
original  recording.  

Rainfall  in  binaural   Good  low  frequency   Higher  frequencies   Recording  made  using  
close.wav   response  to  thunder   much  more  prevalent  in   Harbeth  M20  speakers  
within  the  recordings.   recording.  This   in  stereo  2.0  
The  thunder  also   example  was  not  able   configuration.  
maintains  a  good  sense   to  capture  a  low  end  of  
of  stereo  imagery.     the  rainfall  resulting  in  a  
poor  representation  of  
rain  atmos.    

Rainfall  in  binaural   Good  low  frequency   Similar  to  above   Recording  made  using  
far.wav   response  to  thunder   example,  but  to  a   Harbeth  M20  speakers  
within  the  recordings.   greater  extent.  The   in  stereo  2.0  
The  thunder  also   lower  frequencies  of  the   configuration.  
maintains  a  good  sense   rain  within  the  recording  
of  stereo  imagery.     have  suffered  badly  
due  to  the  worldizing  
technique  
 
Fig  2.  Table  outlining  the  results  of  the  moving  sound  cues  binaural  worldizing  tests.    
 
File   Pros   Cons   Comments  

Long  woosh   Clear  sound,  obvious   Lack  of  externalisation   Original  non  binaural  
pass-­by.wav   panning  from  right  to   when  listened  to  over   audio  used  for  
left.     headphones   playback  in  
worldizing  
experiment.    

Short  woosh    Clear  and  'punchy'   Lack  of  externalisation.   Original  non  binaural  
pass-­by.wav   sound.  Clear  panning   Very  'clinical'  recording   audio  used  for  
from  right  to  left.     of  effect  with  little  to  no   playback  in  
tail.     worldizing  

8
experiment.    

Long  woosh  RR  to   More  externalisation   No  great  definition   Recording  made  using  
RF.wav   than  original  recording,   between  rear  and  front   Harbeth  M20  speakers  
sound  appears  to  be   positions.  Sound   in  stereo  2.0  
further  from  listening   appear  to  originate  from   configuration.  
position,  particularly   a  rear  position,  but   Microphone  
during  the  tail  of  the   does  not  clearly  finnish   arrangement  20cm  
sound  effect.     in  the  opposite  front   from  speakers.  
Frequency  response   position.  Obvious  case  
appears  to  be  mostly   of  front/back  confusion.    
intact  when  compared  
to  original  recording.    

Short  woosh  RR  to   Some  movement  is   Again  this  recording   Recording  made  using  
RF.wav   clear  in  recording,   sufferers  from   Harbeth  M20  speakers  
frequency  response  is   front/back  confusion.     in  stereo  2.0  
good,  also  change  in   configuration.  
dynamics  matches  the   Microphone  
sound  effects  position   arrangement  20cm  
in  space.     from  speakers.  

Long  woosh  L  to   Good  spatial   Obvious  change  in   Recording  made  using  
R.wav   awareness  and  panning   dynamic  range  and   Harbeth  M20  speakers  
from  left  to  right.     amplitude.     in  stereo  2.0  
configuration.  
Microphone  
arrangement  20cm  
from  speakers.  

Short  woosh  L  to   Good  spatial   Less  amplitude  than   Recording  made  using  
R.wav   awareness  and  panning   original  recording.   Harbeth  M20  speakers  
from  left  to  right.   in  stereo  2.0  
Improved   configuration.  
externalisation   Microphone  
compared  to  original   arrangement  20cm  
recording.   from  speakers.  

Long  woosh  Front  to   Good  representation  of   No  obvious  movement   Recording  made  using  
Rear.wav   frequency  and   of  sound  from  front  to   Harbeth  M20  speakers  
dynamics.     rear.  Front/back   in  stereo  2.0  
confusion  obvious  as   configuration.  
recording  moves  from   Microphone  
0​°​  to  180​°.   arrangement  20cm  
from  speakers.  

Short  woosh  Front  to   Good  representation  of   No  obvious  movement   Recording  made  using  
Rear.wav   frequency  and   of  sound  from  front  to   Harbeth  M20  speakers  
dynamics.     rear.  Front/back   in  stereo  2.0  
confusion  obvious  as   configuration.  
recording  moves  from   Microphone  
0​°​  to  180​°.   arrangement  20cm  
from  speakers.  

9
 

8  Conclusion   experiment  may  have  seen  better  results  with  


  with  a  microphone  such  as  the  DPA  4060  that  
The  binaural  worldizing  technique  has  had  both   has  a  frequency  range  of  20Hz-­20kHz.  With  
positive  and  negative  effects  on  the  recordings.   clear  frequency  response  information  from  the  
The  positive  effects  have  included  improved   microphone  manufacturer  it  would  be  easier  to  
spatial  awareness  whilst  keeping  left/right   determine  whether  the  higher  frequencies  of  the  
panning  intact.  Negative  results  have  included   original  sound  effects  could  have  been  recorded  
difference  in  dynamic  range  and  at  times   as  part  of  the  binaural  worldizing,  or  wether  this  
reductions  in  some  frequency  ranges.     cut  off  at  around  15kHz  is  to  be  expected  while  
  using  the  technique.    
 8.1  Speakers  used    
  8.3  Frequency  response  
The  Harbeth  M20s  show  significant    
improvements  over  the  Yamaha  MS101ii.  Both   Throughout  all  of  the  recordings  where  the  
speakers  have  a  good  frequency  response  with   Harbeth  M20s  were  used,  good  frequency  
the  M20s  quoting  a  75Hz-­20kHz,  and  the   response  is  present.  As  stated  previously  the  
MS101ii  quoting  ​75Hz-­18kHz.     Yamaha  MS101ii  gave  less  desirable  results  
The  Harbeths  performed  significantly  better   with  a  lack  of  low  end  response.    
during  the  tests  than  the  Yamaha,  which  is  likely   When  using  the  Harbeth  M20s,  frequency  
down  to  two  factors.  Firstly  the  physical  size  of   response  is  shown  to  be  present  from  around  
the  drivers,  with  the  M20s  110mm  mid/low  and   20Hz-­15kHz.  As  previously  highlighted,  these  
19mm  tweeter,  compared  to  the  MS101iis  single   frequency  contain  a  lot  of  useful  information  for  
100mm  driver.  Secondly  the  enclosure  of  the   both  localisation  and  intelligibility  of  sound  cues.  
speaker,  with  the  M20s  being  constructed  from  a    
wooden  enclosure  and  the  MS101ii  being   8.4  Dynamics  
constructed  of  plastic.    
Both  of  these  factors  contribute  the  the  Harbeths   Dynamic  range  of  the  binaural  recordings  were  
delivering  a  cleaner  sound  truer  to  that  of  the   able  to  be  match  to  the  original  recordings  within  
original  recording,  and  therefore  more  suitable   2dB  with  some  experimentation.  The  research  
for  this  experiment.     found  that  by  placing  the  microphone  array  
  20cm  from  the  speakers,  forming  a  equilateral  
8.2  Microphone  used   triangle  at  normal  listening  height  allowed  the  
  dynamic  range  of  the  recordings  to  stay  within  
The  microphones  used  throughout  the  test  were   this  2dB  range  across  several  speaker  
the  ​LUHD  PM-­01AB  omni-­directional   arrangements.    
microphones,  enclosed  in  the  mould  designed  to    
be  the  size  of  an  average  human  head.  Whilst   8.5  Spatial  perception  /  Room  reflections  
enclosed  on  either  side  of  the  head  mould  the    
stereo  pair  would  have  been  creating  a  binaural   Sound  effects  that  featured  strong  left/right  
effect.     information,  '​short  woosh  pass-­by.wav​'  and  
Unfortunately  there  is  no  frequency  response   '​Rainfall  NOT  BINAURAL.wav​'  for  example,  
specified  by  LUHD  on  their  website,  however   were  able  to  keep  this  information  once  
frequency  response  was  recorded  at  up  to   converted  to  binaural  using  the  worldizing  
12-­15kHz  and  as  low  as  <50Hz.  Within  this   technique.  However,  sound  effects  that  were  
range  both  low  end  frequencies  important  for   worldized  into  a  ​'Front  to  Rear'  ​or  ​'RR  to  RF'  
localisation  [11],  and  higher  frequencies  for   suffered  from  front/back  confusion  [11,12]  and  
intelligibility  of  the  human  voice​  ​[10]​  ​are  left   the  panning  effect  is  lost.    
intact  as  part  of  the  binaural  recording.  The  
 

10
 

All  of  the  binaural  recordings  feature  room   [6]  Costerton,  B  (2013).  A  Systematic  Review  of  
reflections  due  to  the  worldizing  technique.  The   the  Most  Appropriate  Methods  of  Achieving  
recordings  in  this  test  were  all  performed  in  a   Spatially  Enhanced  Audio  for  Headphone  Use.  
deadened  small  vocal  booth,  however  some   Buckinghamshire  New  University.  UK  
room  reflections  exist  in  the  recordings.  These    
room  reflections  appear  to  add  a  sense  of   [7]  Anon  (nd).  PM-­01AB  Stereo  Microphone.  
externalisation  to  the  worldized  sound  effects   [Online]  LUHD.  Available  from:  
which  is  a  positive  quality  for  spatial  audio   https://www.luhd-­mics.com/store/p19/PM-­01AB_
designed  to  be  listened  to  via  headphones.     Stereo_Microphone.html  [Accessed:  20  
  Janurary  2018]    
8.6  Usefulness  of  worldizing  technique    
  [8]  Anon  (nd).  MS101ii.  [Online]  Yamaha.  
Whether  worldized  binaural  sound  effects  would   Available  from:  
be  useful  as  part  of  a  greater  sound  mix  will   http://www.yamahaproaudio.com/global/en/prod
ultimately  be  up  to  the  sound  effects  editor  on   ucts/speakers/ms101iii/  [Accessed  20  Janurary  
any  particular  project.  However  it  is  possible  say   2018]  
that  some  of  the  recordings  produced  as  a  result    
of  the  research  are  of  a  good  enough  quality  in   [9]  Anon  (nd)  ​The  Monitor  20.1  Pro.  [Online]  
spatial  characteristics,  frequency  response,  and   Available  from:  
dynamics  to  be  used  as  part  of  a  sound  mix.     http://www.harbeth.co.uk/studio-­monitors/monito
  r-­20-­1-­pro.php  [Accessed  20  Janurary  2018]  
   
References:     [10]  Purcell,  J.  (2012).  Dialogue  Editing  for  
  Motion  Pictures.  Focal  Press.  Oxford  
[1]  Murch,  W.  (n.d)  Walter  Murch  Articles    
[Online].  Filmsound.org.  Available  from:   [11]  Rumsey,  F.  (2001).  Spatial  Audio.  Focal  
http://filmsound.org/murch/murch.htm   Press.  Oxford  
[Accessed:  20  January  2018]    
  [12]  Satrongar,  D.  Pike,  C.  Lam,  Y.  Tew,  A.  
[2]  Kahn,  A.  (2002),  Kind  of  Blue,  The  Making  of   (2015)  The  Influence  of  Headphones  on  the  
a  masterpiece.  Granta  Publications.  London     Localization  of  External  Loudspeaker  Sources.  
  AES.  USA  
[3]  ​Burrt,  B.  (2001)  Galactic  Phrase  Book  &    
Travel  Guide.  Lucas  Books.  Random  House.    
New  York.    
   
[4]  Anon.  (2014)  Godzilla:    Behind  the  Roar  
[Online].  Soundworks  Collection.  Available  from:  
http://soundworkscollection.com/news/godzilla-­-­
behind-­the-­roar  [Accessed:  20  Janurary  2018]    
 
[5]  Miramontes,  J.  (n.d)  5.1  Re-­recording  
to  Binaural.  [Online]  Stanford  University.  
Available  from:  
https://ccrma.stanford.edu/~jorgem/220a/fi  
nal-­proj.html  [Accessed  20  January  2018]  
 

11

S-ar putea să vă placă și