Users Guide to FlightGear sound configuration Version 0.9.8, October 30, 2005 Author: Erik Hofman This document is an attempt to describe the configuration of FlightGear flight simulator's aircraft sound in XML. Sound Architecture: ------------------ All of the sound configuration files are XML-encoded* property lists. The root element of each file is always named . Tags are almost always found in pairs, with the closing tag having a slash prefixing the tag name, i.e . The exception is the tag representing an aliased property. In this case a slash is prepended to the closing angle bracket. (see section Aliasing) The top level sound configuration file is composed of a , a , a sound file and zero or more and/or definitions. [ Paths are relative to $FG_ROOT (the root of the installed base package .) ] [ Absolute paths may be used. Comments are bracketed with . ] A limited sound configuration file would look something like this: engine Sounds/wasp.wav looped /engines/engine/running /engines/engine/mp-osi 0.005 0.15 0.5 0.15 /engines/engine/rpm 0.0012 0.3 5.0 0.3 This would define an engine sound event handler for a piston engine driven aeroplane. The sound representing the engine is located in $FG_ROOT/Sounds and is named wasp.wav. The event is started when the property /engines/engine/running becomes non zero. When that happens, the sound will be played looped (see ) until the property returns zero again. As you can see the volume is mp-osi dependent, and the pitch of the sound depends on the engine rpm. Configuration description: ------------------------- Named FX subtree living under /sim/sound < ... > This is the event separator. The text inside the brackets can be anything. Bit it is advised to give it a meaningful name like: crank, engine, rumble, gear, squeal, flap, wind or stall The value can be defined multiple times, thus anything which is related may have the same name (grouping them together). This defines the name of the event. This name is used internally and, although it can me defined multiple times in the same file, should normally have an unique value. Multiple definitions of the same name will allow multiple sections to interfere in the starting and stopping of the sample. This method can't be used to control the pitch or volume of the sample, but instead multiple volume or pitch section should be included inside the same event. The types "raise" and "fall" will stop the playback of the sample regardless of any other event. This means that when the type "raise" is supplied, sample playback will stop when the event turns false. Using the type "fall" will stop playback when the event turns true. IMPORTANT: If the trigger is used for anything else but stopping the sound at a certain event, all sections with the same name *should* have exactly the same sections for everything but property and type. In the case of just stopping the sample at a certain event, the sections for path, volume and pitch may be omitted. This defined th path to the sound file. The path is relative to the FlightGear root directory but could be specified absolute. Define a condition that triggers the event. For a complete description of the FlightGear conditions, please read docs-mini/README.conditions An event should define either a condition or a property. Define which property triggers the event, and refers to a node in the FlightGear property tree. Action is taken when the property is non zero. A more sophisticated mechanism to trigger the event is described in This defines how the sample should be played: once: the sample is played once. this is the default. looped: the sample plays continuously, until the event turns false. in-transit: the sample plays continuously, while the property is changing its value. This defines the type os this sample: fx: this is the default type and doesn't need to be defined. avionics: sounds set to this time don't have a position and orientation but are treated as if it's mounted to the aircraft panel. it's up to the user to define if it can always be heard or only when in cockpit view. / Volume or Pitch definition. Currently there may be up to 5 volume and up to 5 pitch definitions defined within one sound event. The volume elements are processed as follows total_offset = 0 total_volume = 1.0 for each element: (a) if an is defined: and the value/result is >= 0 total_volume = total_volume * expression_value process next element (b) Use either or as the base value (c) apply any function to the value (d) value = value * abs(factor) (e) value = max(value, ) (f) value = min(value, ) (g) if was originally negative then use subtract-mode otherwise use normal mode (normal-mode) if (value >= 0) total_offset = total_offset + offset total_volume = total_volume * value (subtract-mode) value = value + offset if (value >= 0) total_volume = total_volume * value Then after processing all of the blocks the total volume will be determined by volume = total_offset + total_volume; If volume exceeds 1 it will be clipped to 1. The pitch elements are processed as follows total_offset = 0 total_pitch = 1.0 for each element: (a) if an is defined: total_pitch = total_pitch * expression_value process next element (b) Use either or as the base value (c) apply any function to the value (d) value = value * abs(factor) (e) value = max(value, ) (f) value = min(value, ) (g) if was originally negative then use subtract-mode otherwise use normal mode (normal-mode) total_offset = total_offset + offset total_pitch = total_pitch * value (subtract-mode) total_pitch = offset - value Then after processing all of the blocks the total pitch will be determined by pitch = total_offset + total_pitch; Defines the AN SGExpression to be used to calculate the volume or pitch. When an expression is used all other tags in the volume block are ignored. Refs: README.expressions http://wiki.flightgear.org/Expressions Property to use for the base value for the calculation. Defines the multiplication factor for the property value. If factor is negative then it will cause the calculation for this element to use the subtract-mode as defined above. The negative value for the factor will be converted to positive The offset elements will be summed and added to the to the calculated volume from all of the volume blocks, except when using subtract-mode in which case the offset is directly added to the calculated volume within each volume block Defines which internal variable should be used for the calculation. The value is treated as a floating point number. The following internals are available at this time: dt_play: the number of seconds since the sound started playing. dt_stop: the number of seconds after the sound has stopped. Defines the delay after the volume block becomes active. An example would be to have two sounds that relate to a single property and the delay-sec element can be used to allow one sound to start after the end of the first one simply by setting the delay-sec to the duration of the sample. Defines the function that should be used upon the property before it is used for calculating the net result: lin: linear handling of the property value. this is the default. ln: convert the property value to a natural logarithmic value before scaling it. Anything below 1 will return zero. log: convert the property value to a true logarithmic value before scaling it. Anything below 1 will return zero. inv: inverse linear handling (1/x). abs: absolute handling of the value (always positive). sqrt: calculate the square root of the absolute value before scaling it. Add a bit of randomness to the offset. Only used for pitch. Minimum allowed value. This is useful if sounds start to sound funny. Anything lower will be truncated to this value. Maximum allowed value. This is useful if sounds gets to loud. Anything higher will be truncated to this value. Specify the position of the sounds source relative to the aircraft center. The coordinate system used is a left hand coordinate system where +Y = left, -Y = right, -Z = down, +Z = up, -X = forward, +X = aft. Distances are in meters. The volume calculation due to distance and orientation of the sounds source ONLY work on mono samples! X dimension offset Y dimension offset Z dimension offset Specify the orientation of the sounds source. The zero vector is default, indicating that a Source is not directional. Specifying a non-zero vector will make the Source directional in the X,Y,Z direction X dimension Y dimension Z dimension The inner edge of the audio cone in degrees (0.0 - 180.0). Any sound withing that angle will be played at the current gain. The outer edge of the audio cone in degrees (0.0 - 180.0). Any sound beyond the outer cone will be played at "outer-gain" volume. The gain at the outer edge of the cone. Set a reference distance of sound in meters. This is the distance where the volume is at its maximum. Volume is clamped to this maximum for any distance below. Volume is attenuated for any distance above. Attenuation depends on reference and maximum distance. See OpenAL specification on "AL_INVERSE_DISTANCE_CLAMPED" mode for details on exact computation. Set the maximum audible distance for the sound in meters. Sound is cut-off above this distance. Known issue: if a sample with mode 'once' goes out of range, it stops playing entirely, and going back into range of the sound does not allow to hear the end of the sample. Creating a configuration file: ------------------------------ To make things easy, there is a default value for most entries to allow a sane configuration when a certain entry is omitted. Default values are: type: lin factor: 1.0 offset: 0.0 for volume, 1.0 for pitch min: 0.0 max: 0.0 (don't check) Calculations are made the following way (for both pitch and volume): value = 0; offs = 0; for (n = 0; n < max; n++) { if (factor < 0) { value += offset[n] - abs(factor[n]) * function(property[n]); } else { value += factor[n] * function(property[n]); offs += offset[n]; } } volume = offs + value; where function can be one of: lin, ln, log, inv, abs or sqrt