Using the folder heirarchy tree to establish these high level groups.

At the top-permost pop-permost sound file level:

  interface
  human
  resource
  attack
  soundscapes
  fauna

Defaults set for these default classses could include:

  GAIN: 1.0
  LOOPING: N 
  PITCH: 1.0
  MIN_GAIN: 0
  MAX_GAIN: 1.0
  CONE_OUTER_GAIN: 1.0
  CONE_INNER_ANGLE: 360
  CONE_OUTER_ANGLE: 360
  PRIORITY: 60
  BUFFER: buffer_*
  VELOCITY: (For Doppler Effects*)
  DIRECTION: ?
  SOURCE_RELATIVE: Y[/code]
  *Which I am very excited to try out, at least in some instances...although probly not many :)

Some examples given the proposed scheme:

Code looks first to determine defaults from audio_class.xml:

  human_footstep_class
   GAIN: 0.5
   LOOPING: N 
   PITCH: 1.0
   MIN_GAIN: 0.0
   MAX_GAIN: 0.5
   CONE_OUTER_GAIN: 1.0
   CONE_INNER_ANGLE: 360
   CONE_OUTER_ANGLE: 360
   PRIORITY: 60
   BUFFER: buffer_01
   VELOCITY: (For Doppler Effects*)
   DIRECTION: ?
   SOURCE_RELATIVE: Y

then to audio_group.xml for soundfile(s) and adjustment properties

  human_footstep_grass
   class=human_footstep_class
   {
   footstep_01.ogg
   footstep_02.ogg
   footstep_03.ogg
   footstep_04.ogg
   footstep_05.ogg
   footstep_06.ogg
   footstep_07.ogg
   footstep_08.ogg
   footstep_09.ogg
   footstep_10.ogg
   footstep_11.ogg
   footstep_12.ogg
   }
  RANDOMIZE FILES IN GROUP: Y
  GAIN RANGE UPPER: 0.5
  GAIN RANGE LOWER: 0.4
  RANDOMIZE GAIN: Y
  PITCH RANGE UPPER: 1.2
  PITCH RANGE LOWER: 0.8
  RANOMIZE PITCH: Y

human_unit actor says:

 walk on grass: use human_footstep_grass from audio_group.xml

or

 cow_footstep_grass
  class=human_footstep_class
  {
  footstep_01
  footstep_02
  footstep_03
  footstep_04
  footstep_05
  footstep_06
  footstep_07
  footstep_08
  footstep_09
  footstep_10
  footstep_11
  footstep_12
  }
   RANDOMIZE FILES IN GROUP: Y
   GAIN RANGE UPPER: 0.5
   GAIN RANGE LOWER: 0.4
   RANDOMIZE GAIN: Y
   PITCH RANGE UPPER: 0.6
   PITCH RANGE LOWER: 0.4
   RANOMIZE PITCH: Y

animal_cow actor says:

 walk on grass: use cow_footstep_grass from audio_group.xml

And with a little deeper implentation:
audio_class.xml

 human_lumbering_class
  GAIN: 1.0
  LOOPING: N 
  PITCH: 1.0
  MIN_GAIN: 0.0
  MAX_GAIN: 1.0
  CONE_OUTER_GAIN: 1.0
  CONE_INNER_ANGLE: 360
  CONE_OUTER_ANGLE: 360
  PRIORITY: 60
  BUFFER: buffer_*
  VELOCITY: (For Doppler Effects*)
  DIRECTION: ?
  SOURCE_RELATIVE: Y

audio_group.xml

 human_lumbering
  class=human_lumbering_class
   {
   chop_01
   chop_02
   chop_03
   chop_04
   chop_05
   chop_06
   chop_07
   chop_08
   }
 RANDOMIZE FILES IN GROUP: Y
 PITCH RANGE UPPER: 1.2
 PITCH RANGE LOWER: 0.8
 RANOMIZE PITCH: Y
  {
  falling_01
  falling_02
  falling_03
  falling_04
  falling_05
  }
   RANDOMIZE FILES IN GROUP: Y
   GAIN RANGE UPPER: 1.0
   GAIN RANGE LOWER: 0.5
   RANDOMIZE GAIN: Y
   PITCH RANGE UPPER: 1.0
   PITCH RANGE LOWER: 0.8
   RANOMIZE PITCH: Y
   INTERMITTENCY: Randomly once every 30-60 seconds

human_lumbering actor says:

 resource gather lumber: use human_lumbering from audio_group.xml

Further Info here:

Managing Audio Parameters