Loudspeaker and listener positions for optimal low-frequency spatial reproduction in listening rooms.

1
Loudspeaker and listener positions for optimal
low-frequency spatial reproduction in listening
rooms.

• David Griesinger Location Lexicon, 3 Oak Park
  Dr., Bedford, MA 01730-1441

2
Main Message(s)

• 1. Low frequency spatiality (envelopment) is
  beautiful.
• 2. We perceive it through fluctuations in the
  apparent azimuth of low frequency signals.
• 3. Fluctuations are produced by the acoustics of
  large spaces, and not by small spaces. We must
  reproduce them from information in the recording.
• 4. A suitable recording must contain
  decorrelated reverberation in at least two
  channels.
• 5. Reproducing the perception of envelopment
  from a recording requires least two full-range
  speakers.
• In anechoic space these speakers should be at the
  sides of the listener
• 6. Perceiving envelopment in a small room
  requires that the pressure gradient at the
  listening position should be independent of the
  pressure.
• And that each should be driven by an independent
  component of the recorded reverberation
• 7. The above conditions are most commonly met
  when pressure gradients produced by assymetric
  lateral room modes overlap with pressure from
  medial modes.
• The overlap must be both in frequency and at the
  location of the listener.
• 8. Room dimensions and loudspeaker placements
  are critical.
• Their effects can be predicted by diagrams of the
  first few room modes.

3
Envelopment is beautiful

• Pictured the Deutches Staatsoper, Berlin
• A Lares system has been in constant use for many
  years.
• The principle benefit of this system is to
  increase low frequency envelopment.
• The Lares system is not perceived by the audience
  as adding to the reverberation.
• The system is considered essential for the
  production of most operas and all ballets.
• With the system on the low frequencies come
  alive, giving the audience a feeling of direct
  involvement with both the music and the emotion
  of the scene.

4
Low frequency envelopment is perceived through
fluctuations in azimuth

• There is not time in this paper to discuss the
  theory of interaural fluctuations. See the
  references. In briefest outline
• Reverberation in large spaces is chaotic in its
  direction of incidence, fluctuating randomly even
  during long notes.
• Large spaces are very seldom in steady-state, so
  the ITD at the listener is in constant state of
  change
• We perceive this fluctuation as envelopment the
  sense of being inside the sound.
• The perception is VERY DIFFERENT from having the
  sound inside the head!
• Reverberation in small spaces does not fluctuate
• because for most musical notes the reverberation
  quickly reaches a steady-state condition.
• To reproduce these fluctuations we must first
  record them
• This is easy.
• Then we must find a way of reproducing them at
  the listening position
• This is hard

5
Reverberation in recordings

• Reverberation in most two-channel recordings is
  independent in the two channels
• Reverberation is frequently recorded by two
  microphones spaced by more than the reverberation
  radius
• Artificial reverberation (which is frequently
  used) is deliberately made such that the two
  output channels are completely independent.

6
Reproducing Fluctuating ITD

• Fluctuating ITDs are easy to reproduce in
  anechoic spaces if the loudspeakers are placed
  at the sides of the listeners
• Since the reverberant sound pressure is
  independent in the two speakers, both the
  pressure and the pressure gradient fluctuate.
• The pressure gradient is produced by the
  difference signal of the two loudspeakers
• The pressure is produced by the sum signal
• The pressure gradient and the pressure are
  independent of each other, so a random ITD is
  produced.

7
Anechoic space - standard stereo

• Standard stereo gives little envelopment in an
  anechoic space because the speakers are not
  lateral.
• The pressure gradient is proportional to the sine
  of the speaker angle

8
Envelopment in small rooms

• Envelopment can be reproduced directly in small
  rooms using the inhomogenious (non-modal)
  component of the room sound
• IF the low frequency loudspeakers are at the
  sides of the listeners!

9
Envelopment with room modes

• To create envelopment with room modes we must use
  the information in the recording to create a
  fluctuating ITD at the listening position.
• This can only be done by separately driving the
  pressure gradient and the pressure at the
  listening position.
• Lets look at examples

10
Large room at HSG at 32Hz

• The large demo room at HSG is 17x23x9.
• The ceiling is highly absorbent.
• The first asymmetric lateral mode is at 32Hz.
• The pressure gradient from this mode (the sound
  velocity) is maximum along the front/back center
  line of the room. This area is shown in green.
• The sound pressure from this mode is a minimum
  along this line. So we do not directly hear this
  mode

11
Large room at HSG at 32Hz and 48Hz

• The front/back mode at 48Hz has a pressure
  maximum at the center of the room. (shown in
  red)
• Note that the 48Hz pressure mode overlaps the
  pressure gradient from the 32Hz mode in the
  center of the room.
• The 32Hz mode is driven by the difference signal
  from the loudspeakers, and the 48Hz mode is
  driven by the sum signal.
• So envelopment will be heard at frequencies
  between 32Hz and 48Hz if the Q of these two modes
  is low.

12
Large room at 96Hz

• There is an asymmetric lateral mode at 96Hz,
  which produces areas of high pressure gradient
  (green)
• There is a front/back mode at 96Hz which produces
  high pressure zones (red)
• Where the two overlap we will have high
  envelopment. (areas shown in blue.)
• Note that in this room envelopment is widespread
  at frequencies near 96Hz.

13
Mixing room at HSG (poor spatiality)

• The mixing room has dimensions of 19x21x9.
  The ceiling is absorbing.
• The room is nearly square (bad idea)
• At 30Hz the assymetric lateral mode produces a
  pressure gradient shown in green.
• The 30Hz front/back mode produces the pressure
  zones shown in red.
• Notice there is NO overlap in the center of the
  room which was the listening position.
• There is thus no envelopment at these frequencies.

14
Mixing room at 84Hz

• There is an asymmetric lateral mode at 87hz, and
  a front back mode at 84Hz.
• There is no envelopment at the center but there
  is envelopment at other positions
• But ONLY if the speakers are in the corners!

15
Mixing room at 84Hz as used

• Alas, the marketing department does not put the
  speakers in the corners.
• The speakers look better when they are closer
  together.
• But this puts them at pressure nulls for the
  asymmetric lateral mode.
• So NO envelopment is produced at these
  frequencies.

16
Mixing room at 55Hz

• There are two modes that overlap at 55Hz A
  symmetric lateral mode, and a symmetric
  front/back mode.
• No envelopment is produced because both modes are
  driven by the sum component of the stereo signal.

17
Mixing room at 55Hz asymmetric listening
position

• If we move the listening position to the left, we
  can produce envelopment at 55Hz.
• Although we are using the same modes as the
  previous slide, now the lateral mode is excited
  by the difference signal in the recording.
• This speaker arrangement also produces some
  envelopment at 30Hz, because the lateral mode is
  driven by the left speaker only.
• This arrangement is one of the best for this room
  but envelopment is still poor compared to the
  Large room.

18
Mixing room best solution

• The best solution for this unfortunate room is to
  put two powerful subwoofers at the sides of the
  listening area, and damp the room modes as
  completely as possible (either with bass traps or
  by electronic equalization.)
• Envelopment will be produced through the
  non-modal component of the room transfer function.

19
Small room the authors studio 32Hz to 48Hz.

• The authors studio is 12x12x9 with a 3x6 bay
  window.
• The 32Hz asymmetric lateral mode produces a
  pressure gradient shown in green.
• The 48Hz front/back mode produces the pressure
  zones shown in red
• There is considerable overlap at the listening
  position, and envelopment is high.

20
Small room at 45-73Hz

• The symmetric lateral mode at 73Hz overlaps with
  a front/back mode at 45Hz.
• Normally this would not produce envelopment, but
  here the left speaker is near the pressure null
  of the lateral mode. So this mode is driven
  primarily by the right speaker only, while the
  pressure mode is driven by the sum of left and
  right.
• Some envelopment is audible at the listening
  position.
• There is also envelopment from the asymmetric
  lateral mode at 105Hz.

Overall this is a very pleasant room for music
mixing or listening. Envelopment seems uniform
and high near the listening position.
21
Small room conventional speaker positions

• All speaker configurations with the speakers
  pointing along the long axis of the room sound
  poor in this space.
• The first asymmetric lateral mode is at 46Hz, and
  the overlap with pressure is poor.
• The next asymmetric lateral mode is at 138Hz,
  which is too high to give a spacious bass.
• Conclusion the speaker positions along the long
  wall have advantages for envelopment because they
  maximize the overlap between the lateral pressure
  gradient and the total pressure.

22
Envelopment in the front/back direction.

• All the discussion so far has assumed the
  listener is facing forward. The assumption is
  seledom true.
• The same arguments can be applied to the
  front/back direction.
• If we want envelopment to be high in both
  directions (which is highly desirable) we need at
  least three LF drivers. (and four is better.)
• We also need at least one additional decorrelated
  source channel. (we can derive this using Logic
  7)
• The large room at HSG is a good example of a room
  which is excellent in both directions. It sounds
  uniformly spacious as you rotate your head.
• Achieving good front/back as well as lateral
  envelopment is one of our goals in automobile
  sound system design
• Automobiles rely on inhomogenious (non-modal)
  sound propigation, so this task is not as
  difficult as it would seem.

23
Reducing the Q of room modes electronically

• The large room at HSG has several modes with high
  Q.
• Of particular interest is the one at 48kHz, which
  adds an obnoxious buzz to bass transients.

24
After electronic equalization

• We can find the exact frequency and Q of these
  modes by examining the waterfall plot.
• We can insert a dip filter with the identical
  frequency and Q.
• The result is electronic damping of the mode.
  Unlike conventional room equalization, this
  damping works everywhere in the room.
• Electronic damping improves the transient
  response, the frequency uniformity of the bass,
  and the envelopment.

25
Conclusions

• The perception of envelopment with sound
  reproduction in small rooms is possible and
  desirable.
• But in many common cases the room dimensions and
  speaker layout conspire to make the effect nearly
  inaudible
• Rooms of approximately square dimension set up
  symmetrically are particularly poor at
  reproducing envelopment.
• In such a room envelopment and the spatial
  properties of bass in general is simply not
  audible.
• These rooms can often be greatly helped by moving
  the subwoofers to the sides of the listening
  position, and by damping the room modes as much
  as possible.
• Lack of envelopment a result of unfavorable room
  dimensions and speaker layout.
• It does not imply that envelopment in small rooms
  cannot exist.
• In many cases a lack of envelopmetnt can be
  corrected by making a diagram of the first few
  room modes, and seeing where they overlap.
• In many cases the lack of envelopment is a direct
  result of putting the main loudspeakers at nulls
  for an important lateral asymmetric mode. Once
  again, putting subwoofers at the sides of the
  listeners, and raising the crossover frequency,
  is a solution.
• Drawing a few pictures of how the asymmetric
  lateral modes and the front/back modes overlap in
  a proposed listening room is a highly worth-while
  exercise.
• If you are stuck with a square room seriously
  consider altering it. You will be glad you did.

26
References

• Griesinger, D. The Psychoacoustics of Apparent
  Source Width, Spaciousness and Envelopment in
  Performance Spaces Acta Acustica Vol. 83 (1997)
  721-731 (preprint available on the authors
  web-page.)
• Griesinger, D. Objective measures of
  spaciousness and envelopment Presented at the
  16th international conference of the AES.
  (preprint available on the authors web-page)
• Griesinger, D. Speaker Placement,
  Externalization, and Envelopment in Home
  Listening Rooms AES preprint 4860 (preprint
  available on the authors web-page.)
• Griesinger, D. General overview of spatial
  impression, envelopment, localization, and
  externalization presented at the 15th
  international conference of the AES, Denmark
  1998. (web page)
• Griesinger, D. Spatial impression and
  Envelopment in Small Rooms AES preprint 4638.
  (web page)