Mercurial > cortex

annotate org/hearing.org @ 255:e6e0bb3057b2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
minor edits
author Robert McIntyre <rlm@mit.edu>
date Mon, 13 Feb 2012 23:44:15 -0700
parents f1a74d23e7e4
children c39b8b29a79e
rev   line source
rlm@162 1 #+title: Simulated Sense of Hearing
rlm@162 2 #+author: Robert McIntyre
rlm@162 3 #+email: rlm@mit.edu
rlm@162 4 #+description: Simulating multiple listeners and the sense of hearing in jMonkeyEngine3
rlm@162 5 #+keywords: simulated hearing, openal, clojure, jMonkeyEngine3, LWJGL, AI
rlm@162 6 #+SETUPFILE: ../../aurellem/org/setup.org
rlm@162 7 #+INCLUDE: ../../aurellem/org/level-0.org
rlm@162 8 #+BABEL: :exports both :noweb yes :cache no :mkdirp yes
rlm@162 9
rlm@162 10 * Hearing
rlm@162 11
rlm@220 12 At the end of this post I will have simulated ears that work the same
rlm@220 13 way as the simulated eyes in the last post. I will be able to place
rlm@220 14 any number of ear-nodes in a blender file, and they will bind to the
rlm@220 15 closest physical object and follow it as it moves around. Each ear
rlm@220 16 will provide access to the sound data it picks up between every frame.
rlm@162 17
rlm@162 18 Hearing is one of the more difficult senses to simulate, because there
rlm@162 19 is less support for obtaining the actual sound data that is processed
rlm@220 20 by jMonkeyEngine3. There is no "split-screen" support for rendering
rlm@220 21 sound from different points of view, and there is no way to directly
rlm@220 22 access the rendered sound data.
rlm@220 23
rlm@220 24 ** Brief Description of jMonkeyEngine's Sound System
rlm@162 25
rlm@162 26 jMonkeyEngine's sound system works as follows:
rlm@162 27
rlm@162 28 - jMonkeyEngine uses the =AppSettings= for the particular application
rlm@162 29 to determine what sort of =AudioRenderer= should be used.
rlm@220 30 - Although some support is provided for multiple AudioRendering
rlm@162 31 backends, jMonkeyEngine at the time of this writing will either
rlm@220 32 pick no =AudioRenderer= at all, or the =LwjglAudioRenderer=.
rlm@162 33 - jMonkeyEngine tries to figure out what sort of system you're
rlm@162 34 running and extracts the appropriate native libraries.
rlm@220 35 - The =LwjglAudioRenderer= uses the [[http://lwjgl.org/][=LWJGL=]] (LightWeight Java Game
rlm@162 36 Library) bindings to interface with a C library called [[http://kcat.strangesoft.net/openal.html][=OpenAL=]]
rlm@220 37 - =OpenAL= renders the 3D sound and feeds the rendered sound directly
rlm@220 38 to any of various sound output devices with which it knows how to
rlm@220 39 communicate.
rlm@162 40
rlm@162 41 A consequence of this is that there's no way to access the actual
rlm@220 42 sound data produced by =OpenAL=. Even worse, =OpenAL= only supports
rlm@220 43 one /listener/ (it renders sound data from only one perspective),
rlm@220 44 which normally isn't a problem for games, but becomes a problem when
rlm@220 45 trying to make multiple AI creatures that can each hear the world from
rlm@220 46 a different perspective.
rlm@162 47
rlm@162 48 To make many AI creatures in jMonkeyEngine that can each hear the
rlm@220 49 world from their own perspective, or to make a single creature with
rlm@220 50 many ears, it is necessary to go all the way back to =OpenAL= and
rlm@220 51 implement support for simulated hearing there.
rlm@162 52
rlm@162 53 * Extending =OpenAL=
rlm@162 54 ** =OpenAL= Devices
rlm@162 55
rlm@162 56 =OpenAL= goes to great lengths to support many different systems, all
rlm@162 57 with different sound capabilities and interfaces. It accomplishes this
rlm@162 58 difficult task by providing code for many different sound backends in
rlm@162 59 pseudo-objects called /Devices/. There's a device for the Linux Open
rlm@162 60 Sound System and the Advanced Linux Sound Architecture, there's one
rlm@162 61 for Direct Sound on Windows, there's even one for Solaris. =OpenAL=
rlm@162 62 solves the problem of platform independence by providing all these
rlm@162 63 Devices.
rlm@162 64
rlm@162 65 Wrapper libraries such as LWJGL are free to examine the system on
rlm@162 66 which they are running and then select an appropriate device for that
rlm@162 67 system.
rlm@162 68
rlm@162 69 There are also a few "special" devices that don't interface with any
rlm@162 70 particular system. These include the Null Device, which doesn't do
rlm@162 71 anything, and the Wave Device, which writes whatever sound it receives
rlm@162 72 to a file, if everything has been set up correctly when configuring
rlm@162 73 =OpenAL=.
rlm@162 74
rlm@162 75 Actual mixing of the sound data happens in the Devices, and they are
rlm@162 76 the only point in the sound rendering process where this data is
rlm@162 77 available.
rlm@162 78
rlm@162 79 Therefore, in order to support multiple listeners, and get the sound
rlm@162 80 data in a form that the AIs can use, it is necessary to create a new
rlm@220 81 Device which supports this features.
rlm@162 82
rlm@162 83 ** The Send Device
rlm@162 84 Adding a device to OpenAL is rather tricky -- there are five separate
rlm@162 85 files in the =OpenAL= source tree that must be modified to do so. I've
rlm@220 86 documented this process [[../../audio-send/html/add-new-device.html][here]] for anyone who is interested.
rlm@162 87
rlm@220 88 Again, my objectives are:
rlm@162 89
rlm@162 90 - Support Multiple Listeners from jMonkeyEngine3
rlm@162 91 - Get access to the rendered sound data for further processing from
rlm@162 92 clojure.
rlm@162 93
rlm@220 94 I named it the "Multiple Audio Send" Deives, or =Send= Device for
rlm@220 95 short, since it sends audio data back to the callig application like
rlm@220 96 an Aux-Send cable on a mixing board.
rlm@220 97
rlm@220 98 Onward to the actual Device!
rlm@220 99
rlm@162 100 ** =send.c=
rlm@162 101
rlm@162 102 ** Header
rlm@162 103 #+name: send-header
rlm@162 104 #+begin_src C
rlm@162 105 #include "config.h"
rlm@162 106 #include <stdlib.h>
rlm@162 107 #include "alMain.h"
rlm@162 108 #include "AL/al.h"
rlm@162 109 #include "AL/alc.h"
rlm@162 110 #include "alSource.h"
rlm@162 111 #include <jni.h>
rlm@162 112
rlm@162 113 //////////////////// Summary
rlm@162 114
rlm@162 115 struct send_data;
rlm@162 116 struct context_data;
rlm@162 117
rlm@162 118 static void addContext(ALCdevice *, ALCcontext *);
rlm@162 119 static void syncContexts(ALCcontext *master, ALCcontext *slave);
rlm@162 120 static void syncSources(ALsource *master, ALsource *slave,
rlm@162 121 ALCcontext *masterCtx, ALCcontext *slaveCtx);
rlm@162 122
rlm@162 123 static void syncSourcei(ALuint master, ALuint slave,
rlm@162 124 ALCcontext *masterCtx, ALCcontext *ctx2, ALenum param);
rlm@162 125 static void syncSourcef(ALuint master, ALuint slave,
rlm@162 126 ALCcontext *masterCtx, ALCcontext *ctx2, ALenum param);
rlm@162 127 static void syncSource3f(ALuint master, ALuint slave,
rlm@162 128 ALCcontext *masterCtx, ALCcontext *ctx2, ALenum param);
rlm@162 129
rlm@162 130 static void swapInContext(ALCdevice *, struct context_data *);
rlm@162 131 static void saveContext(ALCdevice *, struct context_data *);
rlm@162 132 static void limitContext(ALCdevice *, ALCcontext *);
rlm@162 133 static void unLimitContext(ALCdevice *);
rlm@162 134
rlm@162 135 static void init(ALCdevice *);
rlm@162 136 static void renderData(ALCdevice *, int samples);
rlm@162 137
rlm@162 138 #define UNUSED(x) (void)(x)
rlm@162 139 #+end_src
rlm@162 140
rlm@162 141 The main idea behind the Send device is to take advantage of the fact
rlm@162 142 that LWJGL only manages one /context/ when using OpenAL. A /context/
rlm@162 143 is like a container that holds samples and keeps track of where the
rlm@162 144 listener is. In order to support multiple listeners, the Send device
rlm@162 145 identifies the LWJGL context as the master context, and creates any
rlm@162 146 number of slave contexts to represent additional listeners. Every
rlm@162 147 time the device renders sound, it synchronizes every source from the
rlm@162 148 master LWJGL context to the slave contexts. Then, it renders each
rlm@162 149 context separately, using a different listener for each one. The
rlm@162 150 rendered sound is made available via JNI to jMonkeyEngine.
rlm@162 151
rlm@162 152 To recap, the process is:
rlm@162 153 - Set the LWJGL context as "master" in the =init()= method.
rlm@162 154 - Create any number of additional contexts via =addContext()=
rlm@162 155 - At every call to =renderData()= sync the master context with the
rlm@162 156 slave contexts with =syncContexts()=
rlm@162 157 - =syncContexts()= calls =syncSources()= to sync all the sources
rlm@162 158 which are in the master context.
rlm@162 159 - =limitContext()= and =unLimitContext()= make it possible to render
rlm@162 160 only one context at a time.
rlm@162 161
rlm@162 162 ** Necessary State
rlm@162 163 #+name: send-state
rlm@162 164 #+begin_src C
rlm@162 165 //////////////////// State
rlm@162 166
rlm@162 167 typedef struct context_data {
rlm@162 168 ALfloat ClickRemoval[MAXCHANNELS];
rlm@162 169 ALfloat PendingClicks[MAXCHANNELS];
rlm@162 170 ALvoid *renderBuffer;
rlm@162 171 ALCcontext *ctx;
rlm@162 172 } context_data;
rlm@162 173
rlm@162 174 typedef struct send_data {
rlm@162 175 ALuint size;
rlm@162 176 context_data **contexts;
rlm@162 177 ALuint numContexts;
rlm@162 178 ALuint maxContexts;
rlm@162 179 } send_data;
rlm@162 180 #+end_src
rlm@162 181
rlm@162 182 Switching between contexts is not the normal operation of a Device,
rlm@162 183 and one of the problems with doing so is that a Device normally keeps
rlm@162 184 around a few pieces of state such as the =ClickRemoval= array above
rlm@220 185 which will become corrupted if the contexts are not rendered in
rlm@162 186 parallel. The solution is to create a copy of this normally global
rlm@162 187 device state for each context, and copy it back and forth into and out
rlm@162 188 of the actual device state whenever a context is rendered.
rlm@162 189
rlm@162 190 ** Synchronization Macros
rlm@162 191 #+name: sync-macros
rlm@162 192 #+begin_src C
rlm@162 193 //////////////////// Context Creation / Synchronization
rlm@162 194
rlm@162 195 #define _MAKE_SYNC(NAME, INIT_EXPR, GET_EXPR, SET_EXPR) \
rlm@162 196 void NAME (ALuint sourceID1, ALuint sourceID2, \
rlm@162 197 ALCcontext *ctx1, ALCcontext *ctx2, \
rlm@162 198 ALenum param){ \
rlm@162 199 INIT_EXPR; \
rlm@162 200 ALCcontext *current = alcGetCurrentContext(); \
rlm@162 201 alcMakeContextCurrent(ctx1); \
rlm@162 202 GET_EXPR; \
rlm@162 203 alcMakeContextCurrent(ctx2); \
rlm@162 204 SET_EXPR; \
rlm@162 205 alcMakeContextCurrent(current); \
rlm@162 206 }
rlm@162 207
rlm@162 208 #define MAKE_SYNC(NAME, TYPE, GET, SET) \
rlm@162 209 _MAKE_SYNC(NAME, \
rlm@162 210 TYPE value, \
rlm@162 211 GET(sourceID1, param, &value), \
rlm@162 212 SET(sourceID2, param, value))
rlm@162 213
rlm@162 214 #define MAKE_SYNC3(NAME, TYPE, GET, SET) \
rlm@162 215 _MAKE_SYNC(NAME, \
rlm@162 216 TYPE value1; TYPE value2; TYPE value3;, \
rlm@162 217 GET(sourceID1, param, &value1, &value2, &value3), \
rlm@162 218 SET(sourceID2, param, value1, value2, value3))
rlm@162 219
rlm@162 220 MAKE_SYNC( syncSourcei, ALint, alGetSourcei, alSourcei);
rlm@162 221 MAKE_SYNC( syncSourcef, ALfloat, alGetSourcef, alSourcef);
rlm@162 222 MAKE_SYNC3(syncSource3i, ALint, alGetSource3i, alSource3i);
rlm@162 223 MAKE_SYNC3(syncSource3f, ALfloat, alGetSource3f, alSource3f);
rlm@162 224
rlm@162 225 #+end_src
rlm@162 226
rlm@162 227 Setting the state of an =OpenAL= source is done with the =alSourcei=,
rlm@162 228 =alSourcef=, =alSource3i=, and =alSource3f= functions. In order to
rlm@162 229 completely synchronize two sources, it is necessary to use all of
rlm@162 230 them. These macros help to condense the otherwise repetitive
rlm@162 231 synchronization code involving these similar low-level =OpenAL= functions.
rlm@162 232
rlm@162 233 ** Source Synchronization
rlm@162 234 #+name: sync-sources
rlm@162 235 #+begin_src C
rlm@162 236 void syncSources(ALsource *masterSource, ALsource *slaveSource,
rlm@162 237 ALCcontext *masterCtx, ALCcontext *slaveCtx){
rlm@162 238 ALuint master = masterSource->source;
rlm@162 239 ALuint slave = slaveSource->source;
rlm@162 240 ALCcontext *current = alcGetCurrentContext();
rlm@162 241
rlm@162 242 syncSourcef(master,slave,masterCtx,slaveCtx,AL_PITCH);
rlm@162 243 syncSourcef(master,slave,masterCtx,slaveCtx,AL_GAIN);
rlm@162 244 syncSourcef(master,slave,masterCtx,slaveCtx,AL_MAX_DISTANCE);
rlm@162 245 syncSourcef(master,slave,masterCtx,slaveCtx,AL_ROLLOFF_FACTOR);
rlm@162 246 syncSourcef(master,slave,masterCtx,slaveCtx,AL_REFERENCE_DISTANCE);
rlm@162 247 syncSourcef(master,slave,masterCtx,slaveCtx,AL_MIN_GAIN);
rlm@162 248 syncSourcef(master,slave,masterCtx,slaveCtx,AL_MAX_GAIN);
rlm@162 249 syncSourcef(master,slave,masterCtx,slaveCtx,AL_CONE_OUTER_GAIN);
rlm@162 250 syncSourcef(master,slave,masterCtx,slaveCtx,AL_CONE_INNER_ANGLE);
rlm@162 251 syncSourcef(master,slave,masterCtx,slaveCtx,AL_CONE_OUTER_ANGLE);
rlm@162 252 syncSourcef(master,slave,masterCtx,slaveCtx,AL_SEC_OFFSET);
rlm@162 253 syncSourcef(master,slave,masterCtx,slaveCtx,AL_SAMPLE_OFFSET);
rlm@162 254 syncSourcef(master,slave,masterCtx,slaveCtx,AL_BYTE_OFFSET);
rlm@162 255
rlm@162 256 syncSource3f(master,slave,masterCtx,slaveCtx,AL_POSITION);
rlm@162 257 syncSource3f(master,slave,masterCtx,slaveCtx,AL_VELOCITY);
rlm@162 258 syncSource3f(master,slave,masterCtx,slaveCtx,AL_DIRECTION);
rlm@162 259
rlm@162 260 syncSourcei(master,slave,masterCtx,slaveCtx,AL_SOURCE_RELATIVE);
rlm@162 261 syncSourcei(master,slave,masterCtx,slaveCtx,AL_LOOPING);
rlm@162 262
rlm@162 263 alcMakeContextCurrent(masterCtx);
rlm@162 264 ALint source_type;
rlm@162 265 alGetSourcei(master, AL_SOURCE_TYPE, &source_type);
rlm@162 266
rlm@162 267 // Only static sources are currently synchronized!
rlm@162 268 if (AL_STATIC == source_type){
rlm@162 269 ALint master_buffer;
rlm@162 270 ALint slave_buffer;
rlm@162 271 alGetSourcei(master, AL_BUFFER, &master_buffer);
rlm@162 272 alcMakeContextCurrent(slaveCtx);
rlm@162 273 alGetSourcei(slave, AL_BUFFER, &slave_buffer);
rlm@162 274 if (master_buffer != slave_buffer){
rlm@162 275 alSourcei(slave, AL_BUFFER, master_buffer);
rlm@162 276 }
rlm@162 277 }
rlm@162 278
rlm@162 279 // Synchronize the state of the two sources.
rlm@162 280 alcMakeContextCurrent(masterCtx);
rlm@162 281 ALint masterState;
rlm@162 282 ALint slaveState;
rlm@162 283
rlm@162 284 alGetSourcei(master, AL_SOURCE_STATE, &masterState);
rlm@162 285 alcMakeContextCurrent(slaveCtx);
rlm@162 286 alGetSourcei(slave, AL_SOURCE_STATE, &slaveState);
rlm@162 287
rlm@162 288 if (masterState != slaveState){
rlm@162 289 switch (masterState){
rlm@162 290 case AL_INITIAL : alSourceRewind(slave); break;
rlm@162 291 case AL_PLAYING : alSourcePlay(slave); break;
rlm@162 292 case AL_PAUSED : alSourcePause(slave); break;
rlm@162 293 case AL_STOPPED : alSourceStop(slave); break;
rlm@162 294 }
rlm@162 295 }
rlm@162 296 // Restore whatever context was previously active.
rlm@162 297 alcMakeContextCurrent(current);
rlm@162 298 }
rlm@162 299 #+end_src
rlm@162 300 This function is long because it has to exhaustively go through all the
rlm@162 301 possible state that a source can have and make sure that it is the
rlm@162 302 same between the master and slave sources. I'd like to take this
rlm@162 303 moment to salute the [[http://connect.creativelabs.com/openal/Documentation/Forms/AllItems.aspx][=OpenAL= Reference Manual]], which provides a very
rlm@162 304 good description of =OpenAL='s internals.
rlm@162 305
rlm@162 306 ** Context Synchronization
rlm@162 307 #+name: sync-contexts
rlm@162 308 #+begin_src C
rlm@162 309 void syncContexts(ALCcontext *master, ALCcontext *slave){
rlm@162 310 /* If there aren't sufficient sources in slave to mirror
rlm@162 311 the sources in master, create them. */
rlm@162 312 ALCcontext *current = alcGetCurrentContext();
rlm@162 313
rlm@162 314 UIntMap *masterSourceMap = &(master->SourceMap);
rlm@162 315 UIntMap *slaveSourceMap = &(slave->SourceMap);
rlm@162 316 ALuint numMasterSources = masterSourceMap->size;
rlm@162 317 ALuint numSlaveSources = slaveSourceMap->size;
rlm@162 318
rlm@162 319 alcMakeContextCurrent(slave);
rlm@162 320 if (numSlaveSources < numMasterSources){
rlm@162 321 ALuint numMissingSources = numMasterSources - numSlaveSources;
rlm@162 322 ALuint newSources[numMissingSources];
rlm@162 323 alGenSources(numMissingSources, newSources);
rlm@162 324 }
rlm@162 325
rlm@162 326 /* Now, slave is guaranteed to have at least as many sources
rlm@162 327 as master. Sync each source from master to the corresponding
rlm@162 328 source in slave. */
rlm@162 329 int i;
rlm@162 330 for(i = 0; i < masterSourceMap->size; i++){
rlm@162 331 syncSources((ALsource*)masterSourceMap->array[i].value,
rlm@162 332 (ALsource*)slaveSourceMap->array[i].value,
rlm@162 333 master, slave);
rlm@162 334 }
rlm@162 335 alcMakeContextCurrent(current);
rlm@162 336 }
rlm@162 337 #+end_src
rlm@162 338
rlm@162 339 Most of the hard work in Context Synchronization is done in
rlm@162 340 =syncSources()=. The only thing that =syncContexts()= has to worry
rlm@162 341 about is automatically creating new sources whenever a slave context
rlm@162 342 does not have the same number of sources as the master context.
rlm@162 343
rlm@162 344 ** Context Creation
rlm@162 345 #+name: context-creation
rlm@162 346 #+begin_src C
rlm@162 347 static void addContext(ALCdevice *Device, ALCcontext *context){
rlm@162 348 send_data *data = (send_data*)Device->ExtraData;
rlm@162 349 // expand array if necessary
rlm@162 350 if (data->numContexts >= data->maxContexts){
rlm@162 351 ALuint newMaxContexts = data->maxContexts*2 + 1;
rlm@162 352 data->contexts = realloc(data->contexts, newMaxContexts*sizeof(context_data));
rlm@162 353 data->maxContexts = newMaxContexts;
rlm@162 354 }
rlm@162 355 // create context_data and add it to the main array
rlm@162 356 context_data *ctxData;
rlm@162 357 ctxData = (context_data*)calloc(1, sizeof(*ctxData));
rlm@162 358 ctxData->renderBuffer =
rlm@162 359 malloc(BytesFromDevFmt(Device->FmtType) *
rlm@162 360 Device->NumChan * Device->UpdateSize);
rlm@162 361 ctxData->ctx = context;
rlm@162 362
rlm@162 363 data->contexts[data->numContexts] = ctxData;
rlm@162 364 data->numContexts++;
rlm@162 365 }
rlm@162 366 #+end_src
rlm@162 367
rlm@162 368 Here, the slave context is created, and it's data is stored in the
rlm@162 369 device-wide =ExtraData= structure. The =renderBuffer= that is created
rlm@162 370 here is where the rendered sound samples for this slave context will
rlm@162 371 eventually go.
rlm@162 372
rlm@162 373 ** Context Switching
rlm@162 374 #+name: context-switching
rlm@162 375 #+begin_src C
rlm@162 376 //////////////////// Context Switching
rlm@162 377
rlm@162 378 /* A device brings along with it two pieces of state
rlm@162 379 * which have to be swapped in and out with each context.
rlm@162 380 */
rlm@162 381 static void swapInContext(ALCdevice *Device, context_data *ctxData){
rlm@162 382 memcpy(Device->ClickRemoval, ctxData->ClickRemoval, sizeof(ALfloat)*MAXCHANNELS);
rlm@162 383 memcpy(Device->PendingClicks, ctxData->PendingClicks, sizeof(ALfloat)*MAXCHANNELS);
rlm@162 384 }
rlm@162 385
rlm@162 386 static void saveContext(ALCdevice *Device, context_data *ctxData){
rlm@162 387 memcpy(ctxData->ClickRemoval, Device->ClickRemoval, sizeof(ALfloat)*MAXCHANNELS);
rlm@162 388 memcpy(ctxData->PendingClicks, Device->PendingClicks, sizeof(ALfloat)*MAXCHANNELS);
rlm@162 389 }
rlm@162 390
rlm@162 391 static ALCcontext **currentContext;
rlm@162 392 static ALuint currentNumContext;
rlm@162 393
rlm@162 394 /* By default, all contexts are rendered at once for each call to aluMixData.
rlm@162 395 * This function uses the internals of the ALCdevice struct to temporally
rlm@162 396 * cause aluMixData to only render the chosen context.
rlm@162 397 */
rlm@162 398 static void limitContext(ALCdevice *Device, ALCcontext *ctx){
rlm@162 399 currentContext = Device->Contexts;
rlm@162 400 currentNumContext = Device->NumContexts;
rlm@162 401 Device->Contexts = &ctx;
rlm@162 402 Device->NumContexts = 1;
rlm@162 403 }
rlm@162 404
rlm@162 405 static void unLimitContext(ALCdevice *Device){
rlm@162 406 Device->Contexts = currentContext;
rlm@162 407 Device->NumContexts = currentNumContext;
rlm@162 408 }
rlm@162 409 #+end_src
rlm@162 410
rlm@220 411 =OpenAL= normally renders all contexts in parallel, outputting the
rlm@162 412 whole result to the buffer. It does this by iterating over the
rlm@162 413 Device->Contexts array and rendering each context to the buffer in
rlm@162 414 turn. By temporally setting Device->NumContexts to 1 and adjusting
rlm@162 415 the Device's context list to put the desired context-to-be-rendered
rlm@220 416 into position 0, we can get trick =OpenAL= into rendering each context
rlm@220 417 separate from all the others.
rlm@162 418
rlm@162 419 ** Main Device Loop
rlm@162 420 #+name: main-loop
rlm@162 421 #+begin_src C
rlm@162 422 //////////////////// Main Device Loop
rlm@162 423
rlm@162 424 /* Establish the LWJGL context as the master context, which will
rlm@162 425 * be synchronized to all the slave contexts
rlm@162 426 */
rlm@162 427 static void init(ALCdevice *Device){
rlm@162 428 ALCcontext *masterContext = alcGetCurrentContext();
rlm@162 429 addContext(Device, masterContext);
rlm@162 430 }
rlm@162 431
rlm@162 432 static void renderData(ALCdevice *Device, int samples){
rlm@162 433 if(!Device->Connected){return;}
rlm@162 434 send_data *data = (send_data*)Device->ExtraData;
rlm@162 435 ALCcontext *current = alcGetCurrentContext();
rlm@162 436
rlm@162 437 ALuint i;
rlm@162 438 for (i = 1; i < data->numContexts; i++){
rlm@162 439 syncContexts(data->contexts[0]->ctx , data->contexts[i]->ctx);
rlm@162 440 }
rlm@162 441
rlm@162 442 if ((ALuint) samples > Device->UpdateSize){
rlm@162 443 printf("exceeding internal buffer size; dropping samples\n");
rlm@162 444 printf("requested %d; available %d\n", samples, Device->UpdateSize);
rlm@162 445 samples = (int) Device->UpdateSize;
rlm@162 446 }
rlm@162 447
rlm@162 448 for (i = 0; i < data->numContexts; i++){
rlm@162 449 context_data *ctxData = data->contexts[i];
rlm@162 450 ALCcontext *ctx = ctxData->ctx;
rlm@162 451 alcMakeContextCurrent(ctx);
rlm@162 452 limitContext(Device, ctx);
rlm@162 453 swapInContext(Device, ctxData);
rlm@162 454 aluMixData(Device, ctxData->renderBuffer, samples);
rlm@162 455 saveContext(Device, ctxData);
rlm@162 456 unLimitContext(Device);
rlm@162 457 }
rlm@162 458 alcMakeContextCurrent(current);
rlm@162 459 }
rlm@162 460 #+end_src
rlm@162 461
rlm@162 462 The main loop synchronizes the master LWJGL context with all the slave
rlm@220 463 contexts, then iterates through each context, rendering just that
rlm@220 464 context to it's audio-sample storage buffer.
rlm@162 465
rlm@162 466 ** JNI Methods
rlm@162 467
rlm@162 468 At this point, we have the ability to create multiple listeners by
rlm@162 469 using the master/slave context trick, and the rendered audio data is
rlm@162 470 waiting patiently in internal buffers, one for each listener. We need
rlm@162 471 a way to transport this information to Java, and also a way to drive
rlm@162 472 this device from Java. The following JNI interface code is inspired
rlm@220 473 by the LWJGL JNI interface to =OpenAL=.
rlm@162 474
rlm@220 475 *** Stepping the Device
rlm@162 476 #+name: jni-step
rlm@162 477 #+begin_src C
rlm@162 478 //////////////////// JNI Methods
rlm@162 479
rlm@162 480 #include "com_aurellem_send_AudioSend.h"
rlm@162 481
rlm@162 482 /*
rlm@162 483 * Class: com_aurellem_send_AudioSend
rlm@162 484 * Method: nstep
rlm@162 485 * Signature: (JI)V
rlm@162 486 */
rlm@162 487 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_nstep
rlm@162 488 (JNIEnv *env, jclass clazz, jlong device, jint samples){
rlm@162 489 UNUSED(env);UNUSED(clazz);UNUSED(device);
rlm@162 490 renderData((ALCdevice*)((intptr_t)device), samples);
rlm@162 491 }
rlm@162 492 #+end_src
rlm@162 493 This device, unlike most of the other devices in =OpenAL=, does not
rlm@162 494 render sound unless asked. This enables the system to slow down or
rlm@162 495 speed up depending on the needs of the AIs who are using it to
rlm@162 496 listen. If the device tried to render samples in real-time, a
rlm@162 497 complicated AI whose mind takes 100 seconds of computer time to
rlm@162 498 simulate 1 second of AI-time would miss almost all of the sound in
rlm@162 499 its environment.
rlm@162 500
rlm@162 501
rlm@220 502 *** Device->Java Data Transport
rlm@162 503 #+name: jni-get-samples
rlm@162 504 #+begin_src C
rlm@162 505 /*
rlm@162 506 * Class: com_aurellem_send_AudioSend
rlm@162 507 * Method: ngetSamples
rlm@162 508 * Signature: (JLjava/nio/ByteBuffer;III)V
rlm@162 509 */
rlm@162 510 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_ngetSamples
rlm@162 511 (JNIEnv *env, jclass clazz, jlong device, jobject buffer, jint position,
rlm@162 512 jint samples, jint n){
rlm@162 513 UNUSED(clazz);
rlm@162 514
rlm@162 515 ALvoid *buffer_address =
rlm@162 516 ((ALbyte *)(((char*)(*env)->GetDirectBufferAddress(env, buffer)) + position));
rlm@162 517 ALCdevice *recorder = (ALCdevice*) ((intptr_t)device);
rlm@162 518 send_data *data = (send_data*)recorder->ExtraData;
rlm@162 519 if ((ALuint)n > data->numContexts){return;}
rlm@162 520 memcpy(buffer_address, data->contexts[n]->renderBuffer,
rlm@162 521 BytesFromDevFmt(recorder->FmtType) * recorder->NumChan * samples);
rlm@162 522 }
rlm@162 523 #+end_src
rlm@162 524
rlm@162 525 This is the transport layer between C and Java that will eventually
rlm@162 526 allow us to access rendered sound data from clojure.
rlm@162 527
rlm@162 528 *** Listener Management
rlm@162 529
rlm@162 530 =addListener=, =setNthListenerf=, and =setNthListener3f= are
rlm@162 531 necessary to change the properties of any listener other than the
rlm@162 532 master one, since only the listener of the current active context is
rlm@162 533 affected by the normal =OpenAL= listener calls.
rlm@162 534 #+name: listener-manage
rlm@162 535 #+begin_src C
rlm@162 536 /*
rlm@162 537 * Class: com_aurellem_send_AudioSend
rlm@162 538 * Method: naddListener
rlm@162 539 * Signature: (J)V
rlm@162 540 */
rlm@162 541 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_naddListener
rlm@162 542 (JNIEnv *env, jclass clazz, jlong device){
rlm@162 543 UNUSED(env); UNUSED(clazz);
rlm@162 544 //printf("creating new context via naddListener\n");
rlm@162 545 ALCdevice *Device = (ALCdevice*) ((intptr_t)device);
rlm@162 546 ALCcontext *new = alcCreateContext(Device, NULL);
rlm@162 547 addContext(Device, new);
rlm@162 548 }
rlm@162 549
rlm@162 550 /*
rlm@162 551 * Class: com_aurellem_send_AudioSend
rlm@162 552 * Method: nsetNthListener3f
rlm@162 553 * Signature: (IFFFJI)V
rlm@162 554 */
rlm@162 555 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_nsetNthListener3f
rlm@162 556 (JNIEnv *env, jclass clazz, jint param,
rlm@162 557 jfloat v1, jfloat v2, jfloat v3, jlong device, jint contextNum){
rlm@162 558 UNUSED(env);UNUSED(clazz);
rlm@162 559
rlm@162 560 ALCdevice *Device = (ALCdevice*) ((intptr_t)device);
rlm@162 561 send_data *data = (send_data*)Device->ExtraData;
rlm@162 562
rlm@162 563 ALCcontext *current = alcGetCurrentContext();
rlm@162 564 if ((ALuint)contextNum > data->numContexts){return;}
rlm@162 565 alcMakeContextCurrent(data->contexts[contextNum]->ctx);
rlm@162 566 alListener3f(param, v1, v2, v3);
rlm@162 567 alcMakeContextCurrent(current);
rlm@162 568 }
rlm@162 569
rlm@162 570 /*
rlm@162 571 * Class: com_aurellem_send_AudioSend
rlm@162 572 * Method: nsetNthListenerf
rlm@162 573 * Signature: (IFJI)V
rlm@162 574 */
rlm@162 575 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_nsetNthListenerf
rlm@162 576 (JNIEnv *env, jclass clazz, jint param, jfloat v1, jlong device,
rlm@162 577 jint contextNum){
rlm@162 578
rlm@162 579 UNUSED(env);UNUSED(clazz);
rlm@162 580
rlm@162 581 ALCdevice *Device = (ALCdevice*) ((intptr_t)device);
rlm@162 582 send_data *data = (send_data*)Device->ExtraData;
rlm@162 583
rlm@162 584 ALCcontext *current = alcGetCurrentContext();
rlm@162 585 if ((ALuint)contextNum > data->numContexts){return;}
rlm@162 586 alcMakeContextCurrent(data->contexts[contextNum]->ctx);
rlm@162 587 alListenerf(param, v1);
rlm@162 588 alcMakeContextCurrent(current);
rlm@162 589 }
rlm@162 590 #+end_src
rlm@162 591
rlm@162 592 *** Initialization
rlm@162 593 =initDevice= is called from the Java side after LWJGL has created its
rlm@162 594 context, and before any calls to =addListener=. It establishes the
rlm@162 595 LWJGL context as the master context.
rlm@162 596
rlm@162 597 =getAudioFormat= is a convenience function that uses JNI to build up a
rlm@162 598 =javax.sound.sampled.AudioFormat= object from data in the Device. This
rlm@162 599 way, there is no ambiguity about what the bits created by =step= and
rlm@162 600 returned by =getSamples= mean.
rlm@162 601 #+name: jni-init
rlm@162 602 #+begin_src C
rlm@162 603 /*
rlm@162 604 * Class: com_aurellem_send_AudioSend
rlm@162 605 * Method: ninitDevice
rlm@162 606 * Signature: (J)V
rlm@162 607 */
rlm@162 608 JNIEXPORT void JNICALL Java_com_aurellem_send_AudioSend_ninitDevice
rlm@162 609 (JNIEnv *env, jclass clazz, jlong device){
rlm@162 610 UNUSED(env);UNUSED(clazz);
rlm@162 611 ALCdevice *Device = (ALCdevice*) ((intptr_t)device);
rlm@162 612 init(Device);
rlm@162 613 }
rlm@162 614
rlm@162 615 /*
rlm@162 616 * Class: com_aurellem_send_AudioSend
rlm@162 617 * Method: ngetAudioFormat
rlm@162 618 * Signature: (J)Ljavax/sound/sampled/AudioFormat;
rlm@162 619 */
rlm@162 620 JNIEXPORT jobject JNICALL Java_com_aurellem_send_AudioSend_ngetAudioFormat
rlm@162 621 (JNIEnv *env, jclass clazz, jlong device){
rlm@162 622 UNUSED(clazz);
rlm@162 623 jclass AudioFormatClass =
rlm@162 624 (*env)->FindClass(env, "javax/sound/sampled/AudioFormat");
rlm@162 625 jmethodID AudioFormatConstructor =
rlm@162 626 (*env)->GetMethodID(env, AudioFormatClass, "<init>", "(FIIZZ)V");
rlm@162 627
rlm@162 628 ALCdevice *Device = (ALCdevice*) ((intptr_t)device);
rlm@162 629 int isSigned;
rlm@162 630 switch (Device->FmtType)
rlm@162 631 {
rlm@162 632 case DevFmtUByte:
rlm@162 633 case DevFmtUShort: isSigned = 0; break;
rlm@162 634 default : isSigned = 1;
rlm@162 635 }
rlm@162 636 float frequency = Device->Frequency;
rlm@162 637 int bitsPerFrame = (8 * BytesFromDevFmt(Device->FmtType));
rlm@162 638 int channels = Device->NumChan;
rlm@162 639 jobject format = (*env)->
rlm@162 640 NewObject(
rlm@162 641 env,AudioFormatClass,AudioFormatConstructor,
rlm@162 642 frequency,
rlm@162 643 bitsPerFrame,
rlm@162 644 channels,
rlm@162 645 isSigned,
rlm@162 646 0);
rlm@162 647 return format;
rlm@162 648 }
rlm@162 649 #+end_src
rlm@162 650
rlm@220 651 ** Boring Device Management Stuff / Memory Cleanup
rlm@162 652 This code is more-or-less copied verbatim from the other =OpenAL=
rlm@220 653 Devices. It's the basis for =OpenAL='s primitive object system.
rlm@162 654 #+name: device-init
rlm@162 655 #+begin_src C
rlm@162 656 //////////////////// Device Initialization / Management
rlm@162 657
rlm@162 658 static const ALCchar sendDevice[] = "Multiple Audio Send";
rlm@162 659
rlm@162 660 static ALCboolean send_open_playback(ALCdevice *device,
rlm@162 661 const ALCchar *deviceName)
rlm@162 662 {
rlm@162 663 send_data *data;
rlm@162 664 // stop any buffering for stdout, so that I can
rlm@162 665 // see the printf statements in my terminal immediately
rlm@162 666 setbuf(stdout, NULL);
rlm@162 667
rlm@162 668 if(!deviceName)
rlm@162 669 deviceName = sendDevice;
rlm@162 670 else if(strcmp(deviceName, sendDevice) != 0)
rlm@162 671 return ALC_FALSE;
rlm@162 672 data = (send_data*)calloc(1, sizeof(*data));
rlm@162 673 device->szDeviceName = strdup(deviceName);
rlm@162 674 device->ExtraData = data;
rlm@162 675 return ALC_TRUE;
rlm@162 676 }
rlm@162 677
rlm@162 678 static void send_close_playback(ALCdevice *device)
rlm@162 679 {
rlm@162 680 send_data *data = (send_data*)device->ExtraData;
rlm@162 681 alcMakeContextCurrent(NULL);
rlm@162 682 ALuint i;
rlm@162 683 // Destroy all slave contexts. LWJGL will take care of
rlm@162 684 // its own context.
rlm@162 685 for (i = 1; i < data->numContexts; i++){
rlm@162 686 context_data *ctxData = data->contexts[i];
rlm@162 687 alcDestroyContext(ctxData->ctx);
rlm@162 688 free(ctxData->renderBuffer);
rlm@162 689 free(ctxData);
rlm@162 690 }
rlm@162 691 free(data);
rlm@162 692 device->ExtraData = NULL;
rlm@162 693 }
rlm@162 694
rlm@162 695 static ALCboolean send_reset_playback(ALCdevice *device)
rlm@162 696 {
rlm@162 697 SetDefaultWFXChannelOrder(device);
rlm@162 698 return ALC_TRUE;
rlm@162 699 }
rlm@162 700
rlm@162 701 static void send_stop_playback(ALCdevice *Device){
rlm@162 702 UNUSED(Device);
rlm@162 703 }
rlm@162 704
rlm@162 705 static const BackendFuncs send_funcs = {
rlm@162 706 send_open_playback,
rlm@162 707 send_close_playback,
rlm@162 708 send_reset_playback,
rlm@162 709 send_stop_playback,
rlm@162 710 NULL,
rlm@162 711 NULL, /* These would be filled with functions to */
rlm@162 712 NULL, /* handle capturing audio if we we into that */
rlm@162 713 NULL, /* sort of thing... */
rlm@162 714 NULL,
rlm@162 715 NULL
rlm@162 716 };
rlm@162 717
rlm@162 718 ALCboolean alc_send_init(BackendFuncs *func_list){
rlm@162 719 *func_list = send_funcs;
rlm@162 720 return ALC_TRUE;
rlm@162 721 }
rlm@162 722
rlm@162 723 void alc_send_deinit(void){}
rlm@162 724
rlm@162 725 void alc_send_probe(enum DevProbe type)
rlm@162 726 {
rlm@162 727 switch(type)
rlm@162 728 {
rlm@162 729 case DEVICE_PROBE:
rlm@162 730 AppendDeviceList(sendDevice);
rlm@162 731 break;
rlm@162 732 case ALL_DEVICE_PROBE:
rlm@162 733 AppendAllDeviceList(sendDevice);
rlm@162 734 break;
rlm@162 735 case CAPTURE_DEVICE_PROBE:
rlm@162 736 break;
rlm@162 737 }
rlm@162 738 }
rlm@162 739 #+end_src
rlm@162 740
rlm@162 741 * The Java interface, =AudioSend=
rlm@162 742
rlm@162 743 The Java interface to the Send Device follows naturally from the JNI
rlm@220 744 definitions. The only thing here of note is the =deviceID=. This is
rlm@220 745 available from LWJGL, but to only way to get it is with reflection.
rlm@220 746 Unfortunately, there is no other way to control the Send device than
rlm@220 747 to obtain a pointer to it.
rlm@162 748
rlm@220 749 #+include: "../../audio-send/java/src/com/aurellem/send/AudioSend.java" src java
rlm@220 750
rlm@220 751 * The Java Audio Renderer, =AudioSendRenderer=
rlm@220 752
rlm@220 753 #+include: "../../jmeCapture/src/com/aurellem/capture/audio/AudioSendRenderer.java" src java
rlm@220 754
rlm@220 755 The =AudioSendRenderer= is a modified version of the
rlm@220 756 =LwjglAudioRenderer= which implements the =MultiListener= interface to
rlm@220 757 provide access and creation of more than one =Listener= object.
rlm@220 758
rlm@220 759 ** MultiListener.java
rlm@220 760
rlm@220 761 #+include: "../../jmeCapture/src/com/aurellem/capture/audio/MultiListener.java" src java
rlm@220 762
rlm@220 763 ** SoundProcessors are like SceneProcessors
rlm@220 764
rlm@220 765 A =SoundProcessor= is analgous to a =SceneProcessor=. Every frame, the
rlm@220 766 =SoundProcessor= registered with a given =Listener= recieves the
rlm@220 767 rendered sound data and can do whatever processing it wants with it.
rlm@220 768
rlm@220 769 #+include "../../jmeCapture/src/com/aurellem/capture/audio/SoundProcessor.java" src java
rlm@162 770
rlm@162 771 * Finally, Ears in clojure!
rlm@162 772
rlm@220 773 Now that the =C= and =Java= infrastructure is complete, the clojure
rlm@220 774 hearing abstraction is simple and closely parallels the [[./vision.org][vision]]
rlm@220 775 abstraction.
rlm@162 776
rlm@220 777 ** Hearing Pipeline
rlm@162 778
rlm@220 779 All sound rendering is done in the CPU, so =(hearing-pipeline)= is
rlm@220 780 much less complicated than =(vision-pipelie)= The bytes available in
rlm@220 781 the ByteBuffer obtained from the =send= Device have different meanings
rlm@220 782 dependant upon the particular hardware or your system. That is why
rlm@220 783 the =AudioFormat= object is necessary to provide the meaning that the
rlm@220 784 raw bytes lack. =(byteBuffer->pulse-vector)= uses the excellent
rlm@220 785 conversion facilities from [[http://www.tritonus.org/ ][tritonus]] ([[http://tritonus.sourceforge.net/apidoc/org/tritonus/share/sampled/FloatSampleTools.html#byte2floatInterleaved%2528byte%5B%5D,%2520int,%2520float%5B%5D,%2520int,%2520int,%2520javax.sound.sampled.AudioFormat%2529][javadoc]]) to generate a clojure vector of
rlm@220 786 floats which represent the linear PCM encoded waveform of the
rlm@220 787 sound. With linear PCM (pulse code modulation) -1.0 represents maximum
rlm@220 788 rarefaction of the air while 1.0 represents maximum compression of the
rlm@220 789 air at a given instant.
rlm@164 790
rlm@221 791 #+name: hearing-pipeline
rlm@162 792 #+begin_src clojure
rlm@220 793 (in-ns 'cortex.hearing)
rlm@162 794
rlm@220 795 (defn hearing-pipeline
rlm@220 796 "Creates a SoundProcessor which wraps a sound processing
rlm@220 797 continuation function. The continuation is a function that takes
rlm@220 798 [#^ByteBuffer b #^Integer int numSamples #^AudioFormat af ], each of which
rlm@220 799 has already been apprpiately sized."
rlm@162 800 [continuation]
rlm@162 801 (proxy [SoundProcessor] []
rlm@162 802 (cleanup [])
rlm@162 803 (process
rlm@162 804 [#^ByteBuffer audioSamples numSamples #^AudioFormat audioFormat]
rlm@220 805 (continuation audioSamples numSamples audioFormat))))
rlm@162 806
rlm@220 807 (defn byteBuffer->pulse-vector
rlm@220 808 "Extract the sound samples from the byteBuffer as a PCM encoded
rlm@220 809 waveform with values ranging from -1.0 to 1.0 into a vector of
rlm@220 810 floats."
rlm@220 811 [#^ByteBuffer audioSamples numSamples #^AudioFormat audioFormat]
rlm@220 812 (let [num-floats (/ numSamples (.getFrameSize audioFormat))
rlm@220 813 bytes (byte-array numSamples)
rlm@220 814 floats (float-array num-floats)]
rlm@220 815 (.get audioSamples bytes 0 numSamples)
rlm@220 816 (FloatSampleTools/byte2floatInterleaved
rlm@220 817 bytes 0 floats 0 num-floats audioFormat)
rlm@220 818 (vec floats)))
rlm@220 819 #+end_src
rlm@220 820
rlm@220 821 ** Physical Ears
rlm@220 822
rlm@220 823 Together, these three functions define how ears found in a specially
rlm@220 824 prepared blender file will be translated to =Listener= objects in a
rlm@220 825 simulation. =(ears)= extracts all the children of to top level node
rlm@220 826 named "ears". =(add-ear!)= and =(update-listener-velocity!)= use
rlm@220 827 =(bind-sense)= to bind a =Listener= object located at the initial
rlm@220 828 position of an "ear" node to the closest physical object in the
rlm@220 829 creature. That =Listener= will stay in the same orientation to the
rlm@220 830 object with which it is bound, just as the camera in the [[http://aurellem.localhost/cortex/html/sense.html#sec-4-1][sense binding
rlm@220 831 demonstration]]. =OpenAL= simulates the doppler effect for moving
rlm@220 832 listeners, =(update-listener-velocity!)= ensures that this velocity
rlm@220 833 information is always up-to-date.
rlm@220 834
rlm@221 835 #+name: hearing-ears
rlm@220 836 #+begin_src clojure
rlm@164 837 (defvar
rlm@164 838 ^{:arglists '([creature])}
rlm@164 839 ears
rlm@164 840 (sense-nodes "ears")
rlm@164 841 "Return the children of the creature's \"ears\" node.")
rlm@162 842
rlm@163 843 (defn update-listener-velocity!
rlm@162 844 "Update the listener's velocity every update loop."
rlm@162 845 [#^Spatial obj #^Listener lis]
rlm@162 846 (let [old-position (atom (.getLocation lis))]
rlm@162 847 (.addControl
rlm@162 848 obj
rlm@162 849 (proxy [AbstractControl] []
rlm@162 850 (controlUpdate [tpf]
rlm@162 851 (let [new-position (.getLocation lis)]
rlm@162 852 (.setVelocity
rlm@162 853 lis
rlm@162 854 (.mult (.subtract new-position @old-position)
rlm@162 855 (float (/ tpf))))
rlm@162 856 (reset! old-position new-position)))
rlm@162 857 (controlRender [_ _])))))
rlm@162 858
rlm@169 859 (defn add-ear!
rlm@164 860 "Create a Listener centered on the current position of 'ear
rlm@164 861 which follows the closest physical node in 'creature and
rlm@164 862 sends sound data to 'continuation."
rlm@162 863 [#^Application world #^Node creature #^Spatial ear continuation]
rlm@162 864 (let [target (closest-node creature ear)
rlm@162 865 lis (Listener.)
rlm@162 866 audio-renderer (.getAudioRenderer world)
rlm@220 867 sp (hearing-pipeline continuation)]
rlm@162 868 (.setLocation lis (.getWorldTranslation ear))
rlm@162 869 (.setRotation lis (.getWorldRotation ear))
rlm@162 870 (bind-sense target lis)
rlm@163 871 (update-listener-velocity! target lis)
rlm@162 872 (.addListener audio-renderer lis)
rlm@162 873 (.registerSoundProcessor audio-renderer lis sp)))
rlm@220 874 #+end_src
rlm@162 875
rlm@220 876 ** Ear Creation
rlm@220 877
rlm@221 878 #+name: hearing-kernel
rlm@220 879 #+begin_src clojure
rlm@220 880 (defn hearing-kernel
rlm@164 881 "Returns a functon which returns auditory sensory data when called
rlm@164 882 inside a running simulation."
rlm@162 883 [#^Node creature #^Spatial ear]
rlm@164 884 (let [hearing-data (atom [])
rlm@164 885 register-listener!
rlm@164 886 (runonce
rlm@164 887 (fn [#^Application world]
rlm@169 888 (add-ear!
rlm@164 889 world creature ear
rlm@220 890 (comp #(reset! hearing-data %)
rlm@220 891 byteBuffer->pulse-vector))))]
rlm@164 892 (fn [#^Application world]
rlm@164 893 (register-listener! world)
rlm@164 894 (let [data @hearing-data
rlm@164 895 topology
rlm@220 896 (vec (map #(vector % 0) (range 0 (count data))))]
rlm@220 897 [topology data]))))
rlm@164 898
rlm@163 899 (defn hearing!
rlm@164 900 "Endow the creature in a particular world with the sense of
rlm@164 901 hearing. Will return a sequence of functions, one for each ear,
rlm@164 902 which when called will return the auditory data from that ear."
rlm@162 903 [#^Node creature]
rlm@164 904 (for [ear (ears creature)]
rlm@220 905 (hearing-kernel creature ear)))
rlm@220 906 #+end_src
rlm@162 907
rlm@220 908 Each function returned by =(hearing-kernel!)= will register a new
rlm@220 909 =Listener= with the simulation the first time it is called. Each time
rlm@220 910 it is called, the hearing-function will return a vector of linear PCM
rlm@220 911 encoded sound data that was heard since the last frame. The size of
rlm@220 912 this vector is of course determined by the overall framerate of the
rlm@220 913 game. With a constant framerate of 60 frames per second and a sampling
rlm@220 914 frequency of 44,100 samples per second, the vector will have exactly
rlm@220 915 735 elements.
rlm@220 916
rlm@220 917 ** Visualizing Hearing
rlm@220 918
rlm@220 919 This is a simple visualization function which displaye the waveform
rlm@220 920 reported by the simulated sense of hearing. It converts the values
rlm@220 921 reported in the vector returned by the hearing function from the range
rlm@220 922 [-1.0, 1.0] to the range [0 255], converts to integer, and displays
rlm@220 923 the number as a greyscale pixel.
rlm@220 924
rlm@221 925 #+name: hearing-display
rlm@220 926 #+begin_src clojure
rlm@221 927 (in-ns 'cortex.hearing)
rlm@221 928
rlm@189 929 (defn view-hearing
rlm@189 930 "Creates a function which accepts a list of auditory data and
rlm@189 931 display each element of the list to the screen as an image."
rlm@189 932 []
rlm@189 933 (view-sense
rlm@189 934 (fn [[coords sensor-data]]
rlm@220 935 (let [pixel-data
rlm@220 936 (vec
rlm@220 937 (map
rlm@220 938 #(rem (int (* 255 (/ (+ 1 %) 2))) 256)
rlm@220 939 sensor-data))
rlm@220 940 height 50
rlm@221 941 image (BufferedImage. (max 1 (count coords)) height
rlm@189 942 BufferedImage/TYPE_INT_RGB)]
rlm@189 943 (dorun
rlm@189 944 (for [x (range (count coords))]
rlm@189 945 (dorun
rlm@189 946 (for [y (range height)]
rlm@220 947 (let [raw-sensor (pixel-data x)]
rlm@189 948 (.setRGB image x y (gray raw-sensor)))))))
rlm@189 949 image))))
rlm@162 950 #+end_src
rlm@162 951
rlm@220 952 * Testing Hearing
rlm@220 953 ** Advanced Java Example
rlm@220 954
rlm@220 955 I wrote a test case in Java that demonstrates the use of the Java
rlm@220 956 components of this hearing system. It is part of a larger java library
rlm@220 957 to capture perfect Audio from jMonkeyEngine. Some of the clojure
rlm@220 958 constructs above are partially reiterated in the java source file. But
rlm@220 959 first, the video! As far as I know this is the first instance of
rlm@220 960 multiple simulated listeners in a virtual environment using OpenAL.
rlm@220 961
rlm@220 962 #+begin_html
rlm@220 963 <div class="figure">
rlm@220 964 <center>
rlm@220 965 <video controls="controls" width="500">
rlm@220 966 <source src="../video/java-hearing-test.ogg" type="video/ogg"
rlm@220 967 preload="none" poster="../images/aurellem-1280x480.png" />
rlm@220 968 </video>
rlm@220 969 </center>
rlm@224 970 <p>The blue sphere is emitting a constant sound. Each gray box is
rlm@224 971 listening for sound, and will change color from gray to green if it
rlm@220 972 detects sound which is louder than a certain threshold. As the blue
rlm@220 973 sphere travels along the path, it excites each of the cubes in turn.</p>
rlm@220 974 </div>
rlm@220 975 #+end_html
rlm@220 976
rlm@220 977 #+include "../../jmeCapture/src/com/aurellem/capture/examples/Advanced.java" src java
rlm@220 978
rlm@220 979 Here is a small clojure program to drive the java program and make it
rlm@220 980 available as part of my test suite.
rlm@162 981
rlm@221 982 #+name: test-hearing-1
rlm@220 983 #+begin_src clojure
rlm@220 984 (in-ns 'cortex.test.hearing)
rlm@162 985
rlm@220 986 (defn test-java-hearing
rlm@162 987 "Testing hearing:
rlm@162 988 You should see a blue sphere flying around several
rlm@162 989 cubes. As the sphere approaches each cube, it turns
rlm@162 990 green."
rlm@162 991 []
rlm@162 992 (doto (com.aurellem.capture.examples.Advanced.)
rlm@162 993 (.setSettings
rlm@162 994 (doto (AppSettings. true)
rlm@162 995 (.setAudioRenderer "Send")))
rlm@162 996 (.setShowSettings false)
rlm@162 997 (.setPauseOnLostFocus false)))
rlm@162 998 #+end_src
rlm@162 999
rlm@220 1000 ** Adding Hearing to the Worm
rlm@162 1001
rlm@221 1002 To the worm, I add a new node called "ears" with one child which
rlm@221 1003 represents the worm's single ear.
rlm@220 1004
rlm@221 1005 #+attr_html: width=755
rlm@221 1006 #+caption: The Worm with a newly added nodes describing an ear.
rlm@221 1007 [[../images/worm-with-ear.png]]
rlm@221 1008
rlm@221 1009 The node highlighted in yellow it the top-level "ears" node. It's
rlm@221 1010 child, highlighted in orange, represents a the single ear the creature
rlm@221 1011 has. The ear will be localized right above the curved part of the
rlm@221 1012 worm's lower hemispherical region opposite the eye.
rlm@221 1013
rlm@221 1014 The other empty nodes represent the worm's single joint and eye and are
rlm@221 1015 described in [[./body.org][body]] and [[./vision.org][vision]].
rlm@221 1016
rlm@221 1017 #+name: test-hearing-2
rlm@221 1018 #+begin_src clojure
rlm@221 1019 (in-ns 'cortex.test.hearing)
rlm@221 1020
rlm@221 1021 (cortex.import/mega-import-jme3)
rlm@221 1022 (import java.io.File)
rlm@221 1023
rlm@221 1024 (use 'cortex.body)
rlm@221 1025
rlm@221 1026 (defn test-worm-hearing []
rlm@221 1027 (let [the-worm (doto (worm) (body!))
rlm@221 1028 hearing (hearing! the-worm)
rlm@221 1029 hearing-display (view-hearing)
rlm@221 1030
rlm@221 1031 tone (AudioNode. (asset-manager)
rlm@221 1032 "Sounds/pure.wav" false)
rlm@221 1033
rlm@221 1034 hymn (AudioNode. (asset-manager)
rlm@221 1035 "Sounds/ear-and-eye.wav" false)]
rlm@221 1036 (world
rlm@221 1037 (nodify [the-worm (floor)])
rlm@221 1038 (merge standard-debug-controls
rlm@221 1039 {"key-return"
rlm@221 1040 (fn [_ value]
rlm@221 1041 (if value (.play tone)))
rlm@221 1042 "key-l"
rlm@221 1043 (fn [_ value]
rlm@221 1044 (if value (.play hymn)))})
rlm@221 1045 (fn [world]
rlm@221 1046 (light-up-everything world)
rlm@221 1047 (com.aurellem.capture.Capture/captureVideo
rlm@221 1048 world
rlm@221 1049 (File."/home/r/proj/cortex/render/worm-audio/frames"))
rlm@221 1050 (com.aurellem.capture.Capture/captureAudio
rlm@221 1051 world
rlm@221 1052 (File."/home/r/proj/cortex/render/worm-audio/audio.wav")))
rlm@221 1053
rlm@221 1054 (fn [world tpf]
rlm@221 1055 (hearing-display
rlm@221 1056 (map #(% world) hearing)
rlm@221 1057 (File. "/home/r/proj/cortex/render/worm-audio/hearing-data"))))))
rlm@221 1058 #+end_src
rlm@221 1059
rlm@221 1060 In this test, I load the worm with its newly formed ear and let it
rlm@221 1061 hear sounds. The sound the worm is hearing is localized to the origin
rlm@221 1062 of the world, and you can see that as the worm moves farther away from
rlm@221 1063 the origin when it is hit by balls, it hears the sound less intensely.
rlm@221 1064
rlm@221 1065 The sound you hear in the video is from the worm's perspective. Notice
rlm@221 1066 how the pure tone becomes fainter and the visual display of the
rlm@221 1067 auditory data becomes less pronounced as the worm falls farther away
rlm@221 1068 from the source of the sound.
rlm@221 1069
rlm@221 1070 #+begin_html
rlm@221 1071 <div class="figure">
rlm@221 1072 <center>
rlm@221 1073 <video controls="controls" width="600">
rlm@221 1074 <source src="../video/worm-hearing.ogg" type="video/ogg"
rlm@221 1075 preload="none" poster="../images/aurellem-1280x480.png" />
rlm@221 1076 </video>
rlm@221 1077 </center>
rlm@221 1078 <p>The worm can now hear the sound pulses produced from the
rlm@221 1079 hymn. Notice the strikingly different pattern that human speech
rlm@221 1080 makes compared to the insturments. Once the worm is pushed off the
rlm@221 1081 floor, the sound it hears is attenuated, and the display of the
rlm@221 1082 sound it hears beomes fainter. This shows the 3D localization of
rlm@221 1083 sound in this world.</p>
rlm@221 1084 </div>
rlm@221 1085 #+end_html
rlm@221 1086
rlm@221 1087 *** Creating the Ear Video
rlm@221 1088 #+name: magick-3
rlm@221 1089 #+begin_src clojure
rlm@221 1090 (ns cortex.video.magick3
rlm@221 1091 (:import java.io.File)
rlm@221 1092 (:use clojure.contrib.shell-out))
rlm@221 1093
rlm@221 1094 (defn images [path]
rlm@221 1095 (sort (rest (file-seq (File. path)))))
rlm@221 1096
rlm@221 1097 (def base "/home/r/proj/cortex/render/worm-audio/")
rlm@221 1098
rlm@221 1099 (defn pics [file]
rlm@221 1100 (images (str base file)))
rlm@221 1101
rlm@221 1102 (defn combine-images []
rlm@221 1103 (let [main-view (pics "frames")
rlm@221 1104 hearing (pics "hearing-data")
rlm@221 1105 background (repeat 9001 (File. (str base "background.png")))
rlm@221 1106 targets (map
rlm@221 1107 #(File. (str base "out/" (format "%07d.png" %)))
rlm@221 1108 (range 0 (count main-view)))]
rlm@221 1109 (dorun
rlm@221 1110 (pmap
rlm@221 1111 (comp
rlm@221 1112 (fn [[background main-view hearing target]]
rlm@221 1113 (println target)
rlm@221 1114 (sh "convert"
rlm@221 1115 background
rlm@221 1116 main-view "-geometry" "+66+21" "-composite"
rlm@221 1117 hearing "-geometry" "+21+526" "-composite"
rlm@221 1118 target))
rlm@221 1119 (fn [& args] (map #(.getCanonicalPath %) args)))
rlm@221 1120 background main-view hearing targets))))
rlm@221 1121 #+end_src
rlm@221 1122
rlm@221 1123 #+begin_src sh
rlm@221 1124 cd /home/r/proj/cortex/render/worm-audio
rlm@221 1125 ffmpeg -r 60 -i out/%07d.png -i audio.wav \
rlm@221 1126 -b:a 128k -b:v 9001k \
rlm@222 1127 -c:a libvorbis -c:v libtheora worm-hearing.ogg
rlm@221 1128 #+end_src
rlm@220 1129
rlm@220 1130 * Headers
rlm@220 1131
rlm@220 1132 #+name: hearing-header
rlm@220 1133 #+begin_src clojure
rlm@220 1134 (ns cortex.hearing
rlm@220 1135 "Simulate the sense of hearing in jMonkeyEngine3. Enables multiple
rlm@220 1136 listeners at different positions in the same world. Automatically
rlm@220 1137 reads ear-nodes from specially prepared blender files and
rlm@221 1138 instantiates them in the world as simulated ears."
rlm@220 1139 {:author "Robert McIntyre"}
rlm@220 1140 (:use (cortex world util sense))
rlm@220 1141 (:use clojure.contrib.def)
rlm@220 1142 (:import java.nio.ByteBuffer)
rlm@220 1143 (:import java.awt.image.BufferedImage)
rlm@220 1144 (:import org.tritonus.share.sampled.FloatSampleTools)
rlm@220 1145 (:import (com.aurellem.capture.audio
rlm@220 1146 SoundProcessor AudioSendRenderer))
rlm@220 1147 (:import javax.sound.sampled.AudioFormat)
rlm@220 1148 (:import (com.jme3.scene Spatial Node))
rlm@220 1149 (:import com.jme3.audio.Listener)
rlm@220 1150 (:import com.jme3.app.Application)
rlm@220 1151 (:import com.jme3.scene.control.AbstractControl))
rlm@220 1152 #+end_src
rlm@220 1153
rlm@221 1154 #+name: test-header
rlm@220 1155 #+begin_src clojure
rlm@220 1156 (ns cortex.test.hearing
rlm@220 1157 (:use (cortex world util hearing))
rlm@221 1158 (:use cortex.test.body)
rlm@220 1159 (:import (com.jme3.audio AudioNode Listener))
rlm@220 1160 (:import com.jme3.scene.Node
rlm@220 1161 com.jme3.system.AppSettings))
rlm@220 1162 #+end_src
rlm@220 1163
rlm@222 1164 * Source Listing
rlm@222 1165 - [[../src/cortex/hearing.clj][cortex.hearing]]
rlm@222 1166 - [[../src/cortex/test/hearing.clj][cortex.test.hearing]]
rlm@222 1167 #+html: <ul> <li> <a href="../org/hearing.org">This org file</a> </li> </ul>
rlm@222 1168 - [[http://hg.bortreb.com ][source-repository]]
rlm@222 1169
rlm@220 1170 * Next
rlm@222 1171 The worm can see and hear, but it can't feel the world or
rlm@222 1172 itself. Next post, I'll give the worm a [[./touch.org][sense of touch]].
rlm@162 1173
rlm@162 1174
rlm@220 1175
rlm@162 1176 * COMMENT Code Generation
rlm@162 1177
rlm@163 1178 #+begin_src clojure :tangle ../src/cortex/hearing.clj
rlm@220 1179 <<hearing-header>>
rlm@221 1180 <<hearing-pipeline>>
rlm@221 1181 <<hearing-ears>>
rlm@221 1182 <<hearing-kernel>>
rlm@221 1183 <<hearing-display>>
rlm@162 1184 #+end_src
rlm@162 1185
rlm@162 1186 #+begin_src clojure :tangle ../src/cortex/test/hearing.clj
rlm@221 1187 <<test-header>>
rlm@221 1188 <<test-hearing-1>>
rlm@221 1189 <<test-hearing-2>>
rlm@221 1190 #+end_src
rlm@221 1191
rlm@221 1192 #+begin_src clojure :tangle ../src/cortex/video/magick3.clj
rlm@221 1193 <<magick-3>>
rlm@162 1194 #+end_src
rlm@162 1195
rlm@162 1196 #+begin_src C :tangle ../../audio-send/Alc/backends/send.c
rlm@162 1197 <<send-header>>
rlm@162 1198 <<send-state>>
rlm@162 1199 <<sync-macros>>
rlm@162 1200 <<sync-sources>>
rlm@162 1201 <<sync-contexts>>
rlm@162 1202 <<context-creation>>
rlm@162 1203 <<context-switching>>
rlm@162 1204 <<main-loop>>
rlm@162 1205 <<jni-step>>
rlm@162 1206 <<jni-get-samples>>
rlm@162 1207 <<listener-manage>>
rlm@162 1208 <<jni-init>>
rlm@162 1209 <<device-init>>
rlm@162 1210 #+end_src
rlm@162 1211
rlm@162 1212