Spatial Sound Encoding Including Near Field Effect: Introducing Distance Coding Filters and a Viable, New Ambisonic Format

1
Spatial Sound Encoding Including Near Field
EffectIntroducing Distance Coding Filters and a
Viable, New Ambisonic Format

• Jérôme Daniel, France Telecom RD
• jerome.!daniel!_at_!francetelecom.com

2
What for a spatial sound encoding approach

• First Higher Order Ambisonics (HOA)
• Models the acoustic reality of sound field
• Homogeneous directional information
• Scalable multi-channel 3D audio format
• Flexibility (transformations decoding)

• Part I - Fundaments of HOA common conception and
  limitations
• A directional encoding technique (based on
  amplitude panning)
• Only able to handle plane waves, thus artificial
  sound fields
• Part II - Recent improvements
• Distance coding filters able to handle near field
  sources
• New encoding format that supports natural or
  realistic sound fields
• True holophonic rendering (comparison with WFS)
• Efficient DSP tools for positional encoding

3
1st Higher Order Ambisonics fundamentsIntrinsi
c representation properties

• Spherical Harmonic Decomposition Fourier-Bessel
  series

• Sound field represented by coefficients Bmns
• Spherical Harmonic component ? Ambisonic
  Signals
• ?pressure field spatial derivatives of successive
  orders m
• Around a reference point listener point of view

• Intrinsic quality of representation
• Using components Bmns up to a limited order (m?M)
• Angular resolution ? radial expansion wave
  length

4
First Higher Order Ambisonics
FundamentsDirectional encoding

• Approximation of elementary wave fronts as plane
  waves (far sources)
• Spherical harmonic decomposition of a plane wave
  (conveying S)
• ? directional encoding pure amplitude panning

5
First higher order ambisonics
fundamentsDecoding and sound field
reconstruction

• Decoders task to recompose ambisonic sound
  field

Loudspeaker signals S D.B

• With finite distance loudspeakers
• Wave encoded as plane ? reconstructed as
  spherical!
• Sound image is projected over the loudspeaker
  array

6
What about finite distance sources ?

• The case of a spherical wave (point source)
• Incidence (q,d ), distance r, conveying S
• Sph. Harm. Decomposition ? ambisonic components
• ? mathematical encoding equation

Slope m x 6dB/oct

• Near Field Effect
• It affects phase and amplitude ratios between
  spatial derivatives of ? orders
• It models the wave front curvature wave length
• Infinite bass-boost with slope as strong as the
  order is high
• Unstable integrating filters resulting signals
  diverge (esp. for higher orders)

• Consequence on currently adopted HOA encoding
  scheme
• Mathematically powerful but physically unviable
  when dealing with near field sources
• Unable to represent natural or realistic sound
  fields !

7
Summary the past, common way of looking
Ambisonics

• In spite of featuring very attractive properties
• Homogeneous directional representation
• Independent from the rendering loudspeaker layout
• Scalable, flexible
• the commonly accepted definition of HOA suffers
  from limitations
• Pure amplitude panning technique only
  directional encoding
• Plane wave model used for encoded virtual sources
• ? acoustically unrealistic
• Unable to represent natural sound fields (at
  higher orders)
• Neither virtual source encoding nor natural
  recording
• Next
• The key to overcome these limitations
• A viable, modified ambisonic format
• Distance Coding (or Near Field Control) filters
• Other consequences and applications

8
Anticipating a reproduction requirement

• Take into account the finite distance of
  loudspeakers
• Near field compensation is required to correct
  the curvature of loudspeakers waves

Ambisonic signals B
Decoding Matrix D
Reconstructed components Bmns are affected by the
loudspeakers Near-Field Effect Fm(R/c)(w)

• Towards a solution for natural sound field
  representation
• Since compensation of loudspeaker near field is
  required
• why not introducing it from the encoding stage?

9
Distance coding filters viable representation

• Practicable Near Field Coding (NFC) filters
• Model loudspeaker NF compensation virtual
  source NF effect
• at the same time, from the encoding stage

• Positional encoding directional distance
  coding

• Its supports natural/realistic sound fields
• It merely requires a classic matrix decoding
• Implicit parameter reference distance R
  loudspeaker array radius
• Can further adapt to any other loudspeaker
  distance R using NFC filters
• (correct the wave field curvature)

10
Generic efficient DSP tools for encoding

• Design of digital NFC (distance coding) filters
• Parametric, minimal-cost IIR filters
• Filter coefficients functions of r, R, c, fs,
  and tabulated roots

• Computation of directional encoding gains
  Ymns(q,d)
• Efficient, recursive algorithm
• Virtually unlimited order
• A complete positional coding scheme

11
Illustration of sound field reconstruction
Frequency domain
Time domain
Gaussian-modulated sine pulse (fc 500 Hz)

• Outside virtual source
• Quite efficient reconstruction
• Better than Wave Field Synthesis
• Daniel et al, AES114
• Becomes easier when virtual source gets closer to
  the real ones
• Enclosed virtual source
• Full reconstruction physically impossible anyway
• Large amount of energy at low frequencies
• Strong interference beyond the virtual source
  distance
• Right direction of propagation (?? WFS)

12
Other consequences and applications

• HOA recording systems become practicable
• by introducing NF compensation at the stage of
  the mic signal processing
• otherwise equalization filters are unstable
  Daniel et al, AES114 and natural sound fields
  cannot be physically represented
• 4th order microphone prototype being
  experimented at FTRD Labs
• Accurate binaural synthesis of close sources
• Virtual Ambisonics
• head-centred ambisonics binaural synthesis of
  virtual loudspeakers
• Supports sound field rotations, thus
  head-tracking adaptation
• Ear-centred double virtual ambisonics
• ?Binaural B-format encoding scheme enriched by
  NF-Coding filters
• More accurate than head-centred virtual
  ambisonics
• but doesnt support rotations once the encoding
  is done

13
Conclusions

• Recall of some HOA fundaments and properties
• Powerful mathematical approach
• Highly versatile 3D sound field representation
• New positional coding completed by distance
  coding
• Efficient digital Near Field Control filters
• Take care of the special case of enclosed sources
• A New, Viable HOA format NFC HOA
• It enables all HOA nice features considering
  natural/realistic sound fields
• It doesnt necessarily interest only systems with
  very numerous loudspeakers

14
Work in progress

• Evaluation of holophonic sound imaging over 48
  loudspeakers
• Up to 15th order Ambisonics
• Comparison of HOA and WFS for outside and
  enclosed sources
• In the context of the CARROUSO project
• Experimentation of a 4th order ambisonic
  microphone
• 32 capsules over a sphere
• Specification of a generic HOA format
• In MPEG-4 (for AudioBIFS V3)
• gt handle multi-channel audio streams as scalable
  3D sound fields
• As a extension of the Wave file format
• Detailed propositions in the paper