Transcript Lecture 4
Audio Engine Programming
Lecture 4
Introduction to FMOD programmers
API
3D Sound
Sounds have position and velocity
There is a listener component
Relationship between the two
Attenuation (with distance)
Occlusion (low-pass filter)
Doppler (relative velocities)
Lots of “psycho-acoustic” options
Fmod audio development
There are three primary components:
The Sound Designer
The Sandbox
The API
We’ll use all three
Start with Sound Designer
Data Driven vs. Low Level API
Low-Level API
Everything is done using programming
The programmer has complete control
The benefits are lower memory overhead
Continued…
Data-Driven API
Takes assets from FMOD Designer
Sound Designer has control over the audio
Integrates them into the code using the
EventSystem object
Key Features
The data-driven approach of the
EventSystem:
Support’s event logic
Allows the Sound Designer to fine-tune or ‘tweak’ the
final mix without the need for Programmer intervention
Creates a logical separate of concerns for the
Programmer and Sound Designer
Overview of event model
FEV
event data
FSB
audio data
FMOD EX
FMOD EventSystem
FMOD Low Level
Assets from the sound designer
The build process creates the following assets:
.FEV file - contains the event meta data
.FSB file(s) - contains the audio data
The ‘Programmer report ‘- a human readable list of
event names, sound definitions [Optional]
A ‘header’ file - A source code header file [Optional]
Cache files - Files used to speed up subsequent
builds (These files are not required by the
Programmer)
Linking the library
Include the EventSystem header file
Declare a pointer to an EventSystem
FMOD::EventSystem *eventsystem;
Create the object (allocate memory)
#include "fmod_event.hpp“
FMOD::EventSystem_Create(&eventsystem);
Initialize the object
eventsystem->init(64, FMOD_INIT_NORMAL, 0,
FMOD_EVENT_INIT_NORMAL);
Loading files
Set the media path (location of files)
Load the .FEV file
eventsystem->load("examples.fev", 0, 0);
Load an event group
eventsystem->setMediaPath((char *)MEDIA_PATH);
eventsystem->getGroup("examples/examples", FMOD_EVENT_DEFAULT,
&eventgroup);
Load an individual event
eventgroup->getEvent(“car", FMOD_EVENT_DEFAULT, &car);
Loading Audio
The required .FSB file(s) are listed in .FEV file
Both the .FEV and .FSB files should be located in the
directory specified with setMediaPath()
The location of audio within the .FSB files is
stored in the .FEV
FMOD will load the audio data from the .FSB
files as required for an event, automatically
Loading Options
FMOD_EVENT_DEFAULT
Specifies default loading behaviour
Event data for the whole group is NOT cached
Loading will block any other task until complete.
FMOD_EVENT_NONBLOCKING
FMOD will use a thread to load the data
Use Event::getState to find out when loading is
complete.
Continued...
FMOD_EVENT_INFOONLY
Does not allocate instances or load data
Creates a ‘handle’ that allows the programmer to
get information from the event
EVENT Playback
Each events have a number of methods and
attributes
To start playback of an event:
example->start();
To stop an event:
Example->stop();
Updating
Updating the event system should be called
once per 'game' tick (or once per frame)
eventsystem->update();
This allows the FMOD Engine to refresh data
such as:
the position of 3D sounds
changes to volume, pitch, effects
as well as internal mechanisms
Demonstration
Simple Event
Demonstrates the necessary code to start
playback of an event
PROVIDING Parameter DATA
Get the parameter called ‘test’
To get the range of the parameter
example->getParameter(“test", &test);
test will be a EventParameter object
test->getRange(&rangemin, &rangemax);
To set the value of the parameter
test->setValue(value);
Demonstration
Parameter
Demonstrates the necessary code to provide
parameter data to an event
PROVIDING a KEY-OFF
If an event parameter is currently sustaining
on a ‘sustain point’, triggering a keyoff will
release it and allow the parameter to continue
test->keyoff()
Demonstration
Keyoff
Demonstrating the necessary code to leave a
sustain point
Selecting sounds at run-time
The sound to be played can be selected at
runtime
This is required for situations such as ‘colour
commentary’ in sport games or other situations
that require dynamic dialogue
The are two methods:
Programmer Sounds
Programmer Selected
Demonstration
Programmer Sound
When a sound definition containing a ‘ProgrammerSound’ is to
be played, a callback is made
The code must then provide a sound object at runtime just
before it's played
Demonstrates the necessary code to select a
sound to play at runtime, using the
‘ProgrammerSound’ method
Demonstration
Programmer Selected
When a sound definition with the playmode
‘ProgrammerSelected’ is to be played, a callback is made
The code must then provide the index number at runtime just
before it's played
The index number identifies which file within the sound definition
is to be played
Demonstrates the necessary code to select a
sound to play at runtime, using the
‘ProgrammerSelected’ method
Closing the Event System
Unload event data
Release audio data in memory
eventsystem->unload();
fsb->release();
Release memory and close eventsystem
memory
eventsystem->release();
Setting up a Project
Take the advice of the video tutorials
Have a separate folder
Copy sounds into a “sounds” directory
Keeps the safe
Can have sub-directories
Create an “output” directory
Designer Interface
Can delete this if you want, but need to have at
least one group
Events
Used to define sound properties for an event
Can be comprised of
One sound
Several sounds randomly chosen
Monophonic vs Polyphonic
Can randomize pitch and attenuation
Events can be
Simple
Multi-track
A Simple Event
Granular Sounds
These sounds aren’t
looping
Allows for a sounds to
occur between time
ranges
Allows for polyphony
Event Options
Can vary attenuation
(dropoff with distance)
Can vary pitch
Creates a sound
considerably smaller
than a “soundtrack”
Plays forever and is
always random!
Sound Definitions
Required for multi-track events
Multi-track Events
Comprised of sound defs in layers
Multi-track Events
Careful when adding sounds
Choose sound def
Ambient noise is looping
Other sounds are granular,
so choose “Oneshot”
Have parameters
Not based on time!
Multi-track Events
Can cross-fade and set fade out time
Effects
Parameter Properties
Can define range and velocity (to simulate
time)
Engine Designer
fmod can specifically work with engine
sounds
Based on the “load” of the engine
Right-click on a
sound instance->properties
Need for Speed 2
Auto-pitch
Window->fmod Engine Designer
The Build Process
Know which platform you’re targeting
Changes are applied only to that platform
Project->Clean, Project->Build
Interactive Music
Comprised of several short segments of
music
Intro music
Darkening or discovery
Fighting/intense fighting
Release
Before you begin, you must know
Tempo of music (beats per minute – bpm)
Beats per measure (time signature)
Cues
Themes and Auditioning
Transitioning via Parameters
Other Things
Supports deployment of multiple languages
Can deploy different builds based on
Platform
Language
A Look at the API
Basic API
Designer API
Written for C/C++
Built on basic API
Can read .fev files
#include <fmod.h> // C
#include <fmod.hpp>
// C++
Read the documentation fmodex.chm
API Parts
A System is the fmod engine
A Sound is the raw resource
2D uses FMOD_2D
3D uses FMOD_3D
result = system->createSound(“music.wav", FMOD_2D, 0, &sound);
API Parts
A Channel is an instance of a Sound
Each time you play a sound, you get a new
channel
Channels can start out paused
You can set the
Volume (0.0f – 1.0f)
Pan (-1.0f – 1.0f)
Frequency (in Hz as a float)
Always use FMOD_CHANNEL_FREE to pick for you
#include <iostream>
#include <fmod.hpp>
#include <fmod_errors.h>
using namespace std;
using namespace FMOD;
void main () {
FMOD_RESULT result;
System* system;
// Create the fmod system. We only need one of these
result = System_Create(&system);
// Specify a max number of simultaneous channels
result = system->init(100, FMOD_INIT_NORMAL, 0);
Sound* sound;
// Decompress the entire mp3 into 16-bit PCM in memory
result = system->createSound("winning.mp3", FMOD_DEFAULT, 0, &sound);
if (result != FMOD_OK) {
cout << "Couldn't open it! " <<
}
FMOD_ErrorString(result) << endl;
Channel* channel;
// Used for setting volume, pan, pausing...
// You have the option of passing that channel as the last parameter
result = system->playSound(FMOD_CHANNEL_FREE, sound, false, &channel);
channel->setFrequency (44100.0f);
if (result != FMOD_OK) {
cout << "Couldn't play it! " << FMOD_ErrorString(result) << endl;
}
while (true) {
cout << "1"; system->update(); // Now required in fmod
}
}
Virtual Voices
Supports “virtual voices” when hardware is
limited
Set from System::init()
System::getCPUUsage()
Try to keep < 1000
May need to set the
priority of a channel
if it’s important
channel->isVirtual()
3D Sound in fmod
Sound attenuation
Logarithmic
Set the mindistance of a sound channel to start
attenuation
Bee = 0.1f
Jet = 50.0f
Leave max distance alone (default is 10,000)
Set 3D settings with System::set3DSettings()
Doppler
Distance factor (in cm, m, feet)
Rolloff scale (attenuation models)
3D Sound
fmod uses a left-handed coordinate system
result=system->init(100,
(FMOD_MODE)(FMOD_INIT_3D_RIGHTHANDED|FMOD_3D), 0);
With velocity, you must pass it as metres per
second
velx = (posx-lastposx) * 1000 / timedelta;
vel = 0.1 * 1000 / 16.67 = 6 meters per second
vel = 0.2 * 1000 / 33.33 = 6 meters per second
void main () {
FMOD_RESULT result;
FMOD_VECTOR soundPos, soundVel;
FMOD_VECTOR listenerPos, listenerVel, listenerForward, listenerUp;
System* system;
result = System_Create(&system);
result = system->init(100, (FMOD_MODE)(FMOD_3D), 0);
int numHC = 0;
result = system->getHardwareChannels(&numHC);
cout << "Hardware channels: " << numHC << endl;
Sound* sound;
result = system->createSound(“train.mp3", FMOD_3D, 0, &sound);
if (result != FMOD_OK) {
cout << "Couldn't open it! " << FMOD_ErrorString(result) << endl;
}
Channel* channel;
result = system->playSound(FMOD_CHANNEL_FREE, sound, false, &channel);
channel->setFrequency(44100.0f);
if (result != FMOD_OK) {
cout << "Couldn't play it! " << FMOD_ErrorString(result) << endl;
}
float dsp, stream, geometry, update, total;
soundPos.y = soundVel.x = soundVel.y = 0.0f;
soundPos.z = -100.0f; soundPos.x = 5.0f;
soundVel.z = 6.0f;
channel->set3DMinMaxDistance(10, 10000);
listenerPos.x = listenerPos.y = listenerPos.z = 0.0f;
listenerVel.x = listenerVel.y = listenerVel.z = 0.0f;
listenerForward.x = listenerForward.y = listenerUp.x = listenerUp.z = 0.0f;
listenerForward.z = listenerUp.y = 1.0f;
system->set3DListenerAttributes(0, &listenerPos, &listenerVel, &listenerForward, &listenerUp);
while (true) {
system->update();
system->getCPUUsage(&dsp, &stream, &geometry, &update, &total);
cout << total << endl;
soundPos.z+=0.01f;
channel->set3DAttributes(&soundPos, &soundVel);
}
}
Interfacing with Sound
Designer
You should have created
.fev file (Designer file – event data)
.fsb file (raw audio data)
.txt file (describes the events and parameters)
May want to start by creating a helper function
void checkErrors(FMOD_RESULT result) {
if (result != FMOD_OK) {
cout << "fmod error: " << FMOD_ErrorString(result) << endl;
exit(1);
}
}
Imagine That This is our
Project
#include <iostream>
#include <fmod.hpp>
#include <fmod_event.hpp>
#include <fmod_errors.h>
using namespace std;
using namespace FMOD;
void main () {
EventSystem* eventSystem = NULL;
EventGroup* eventGroup = NULL;
Event* myEvent = NULL;
EventParameter* eventPar;
FMOD_RESULT result = EventSystem_Create(&eventSystem);
checkErrors(result);
result = eventSystem->init(64, FMOD_INIT_NORMAL, 0, FMOD_EVENT_INIT_NORMAL);
eventSystem->setMediaPath("..//");
result = eventSystem->load("fmodTest.fev", 0, 0);
result = eventSystem->getGroup("fmodTest/beeps", false, &eventGroup);
result = eventGroup->getEvent("PossessedComputer", FMOD_EVENT_DEFAULT, &myEvent);
result = myEvent->getParameter("proximityToComputer", &eventPar);
myEvent->start();
float dir = 0.0001f;
float currentParVal = -1.0f;
eventPar->setValue(0.1f);
while (true){
cout << currentParVal << endl;
eventPar->getValue(¤tParVal);
currentParVal+=dir;
eventPar->setValue(currentParVal);
if ((currentParVal >= 1.0)||(currentParVal <= 0.0)) {
dir = -dir;
}
}
}
Interactive Music
void main () {
EventSystem* eventSystem = NULL;
MusicSystem* musicSystem = NULL;
MusicPrompt* introPrompt, *fightPrompt, *fight2Prompt, *releasePrompt;
FMOD_MUSIC_ITERATOR cueIter;
FMOD_MUSIC_ITERATOR paramIter;
FMOD_MUSIC_ID intensityParID = -1.0f;
FMOD_RESULT result = EventSystem_Create(&eventSystem);
result = eventSystem->init(64, FMOD_INIT_NORMAL, 0,
FMOD_EVENT_INIT_NORMAL);
result = eventSystem->load("acid_test.fev", 0, 0);
result = eventSystem->getMusicSystem(&musicSystem);
musicSystem->setVolume(1.0f);
result = musicSystem->getCues(&cueIter);
cout << "Iter ID: " << cueIter.value->id << endl;
result = musicSystem->prepareCue(cueIter.value->id, &fightPrompt);
result = musicSystem->getNextCue(&cueIter);
cout << "Iter ID: " << cueIter.value->id << endl;
result = musicSystem->prepareCue(cueIter.value->id, &fight2Prompt);
result = musicSystem->getNextCue(&cueIter);
result = musicSystem->prepareCue(cueIter.value->id, &introPrompt);
result = musicSystem->getNextCue(&cueIter);
result = musicSystem->prepareCue(cueIter.value->id, &releasePrompt);
musicSystem->getParameters(¶mIter);
intensityParID = paramIter.value->id;
cout << "paramIter.value->name is " << paramIter.value->name << endl;
musicSystem->setParameterValue(intensityParID, 7.5f);
result = introPrompt->begin();
}
while (true){
eventSystem->update();
}
Other Things You Should Look
Into…
The 3D Reverb API
Asynchronously loading data (by default)
Memory management for non-PC platforms
FMOD::Memory_Initialize() // fix size of fmod
Using compressed samples
Built-in DSP
Configuration Notes
You’ll be working with
Header files
Lib files
C/C++ -> General->Additional Include Directories
Linker->General->Additional Library Directories
Linker->General->Additional Dependencies
DLLs (which you should put into
Windows/System32)
There’s also a FMOD.NET site…
Configuration Notes
You have the option of 64-bit and 32-bit
libraries
You might have to create a new platform