Generic Speech Recognition API. More...
Go to the source code of this file.
Data Structures | |
struct | ast_speech |
struct | ast_speech_engine |
struct | ast_speech_result |
Enumerations | |
enum | ast_speech_flags { AST_SPEECH_QUIET = (1 << 0), AST_SPEECH_SPOKE = (1 << 1), AST_SPEECH_HAVE_RESULTS = (1 << 2) } |
enum | ast_speech_results_type { AST_SPEECH_RESULTS_TYPE_NORMAL = 0, AST_SPEECH_RESULTS_TYPE_NBEST } |
enum | ast_speech_states { AST_SPEECH_STATE_NOT_READY = 0, AST_SPEECH_STATE_READY, AST_SPEECH_STATE_WAIT, AST_SPEECH_STATE_DONE } |
Functions | |
int | ast_speech_change (struct ast_speech *speech, const char *name, const char *value) |
Change an engine specific attribute. | |
int | ast_speech_change_results_type (struct ast_speech *speech, enum ast_speech_results_type results_type) |
Change the type of results we want. | |
int | ast_speech_change_state (struct ast_speech *speech, int state) |
Change state of a speech structure. | |
int | ast_speech_destroy (struct ast_speech *speech) |
Destroy a speech structure. | |
int | ast_speech_dtmf (struct ast_speech *speech, const char *dtmf) |
Signal to the engine that DTMF was received. | |
int | ast_speech_grammar_activate (struct ast_speech *speech, const char *grammar_name) |
Activate a grammar on a speech structure. | |
int | ast_speech_grammar_deactivate (struct ast_speech *speech, const char *grammar_name) |
Deactivate a grammar on a speech structure. | |
int | ast_speech_grammar_load (struct ast_speech *speech, const char *grammar_name, const char *grammar) |
Load a grammar on a speech structure (not globally). | |
int | ast_speech_grammar_unload (struct ast_speech *speech, const char *grammar_name) |
Unload a grammar. | |
struct ast_speech * | ast_speech_new (const char *engine_name, int formats) |
Create a new speech structure. | |
int | ast_speech_register (struct ast_speech_engine *engine) |
Register a speech recognition engine. | |
int | ast_speech_results_free (struct ast_speech_result *result) |
Free a set of results. | |
struct ast_speech_result * | ast_speech_results_get (struct ast_speech *speech) |
Get speech recognition results. | |
void | ast_speech_start (struct ast_speech *speech) |
Indicate to the speech engine that audio is now going to start being written. | |
int | ast_speech_unregister (const char *engine_name) |
Unregister a speech recognition engine. | |
int | ast_speech_write (struct ast_speech *speech, void *data, int len) |
Write audio to the speech engine. |
Generic Speech Recognition API.
Definition in file speech.h.
enum ast_speech_flags |
Definition at line 31 of file speech.h.
00031 { 00032 AST_SPEECH_QUIET = (1 << 0), /* Quiet down output... they are talking */ 00033 AST_SPEECH_SPOKE = (1 << 1), /* Speaker spoke! */ 00034 AST_SPEECH_HAVE_RESULTS = (1 << 2), /* Results are present */ 00035 };
Definition at line 45 of file speech.h.
00045 { 00046 AST_SPEECH_RESULTS_TYPE_NORMAL = 0, 00047 AST_SPEECH_RESULTS_TYPE_NBEST, 00048 };
enum ast_speech_states |
AST_SPEECH_STATE_NOT_READY | |
AST_SPEECH_STATE_READY | |
AST_SPEECH_STATE_WAIT | |
AST_SPEECH_STATE_DONE |
Definition at line 38 of file speech.h.
00038 { 00039 AST_SPEECH_STATE_NOT_READY = 0, /* Not ready to accept audio */ 00040 AST_SPEECH_STATE_READY, /* Accepting audio */ 00041 AST_SPEECH_STATE_WAIT, /* Wait for results to become available */ 00042 AST_SPEECH_STATE_DONE, /* Processing is all done */ 00043 };
int ast_speech_change | ( | struct ast_speech * | speech, | |
const char * | name, | |||
const char * | value | |||
) |
Change an engine specific attribute.
Definition at line 170 of file res_speech.c.
References ast_speech_engine::change, and ast_speech::engine.
Referenced by handle_speechset(), and speech_engine_write().
int ast_speech_change_results_type | ( | struct ast_speech * | speech, | |
enum ast_speech_results_type | results_type | |||
) |
Change the type of results we want.
Definition at line 264 of file res_speech.c.
References ast_speech_engine::change_results_type, ast_speech::engine, and ast_speech::results_type.
Referenced by speech_results_type_write().
00265 { 00266 speech->results_type = results_type; 00267 00268 return (speech->engine->change_results_type ? speech->engine->change_results_type(speech, results_type) : 0); 00269 }
int ast_speech_change_state | ( | struct ast_speech * | speech, | |
int | state | |||
) |
Change state of a speech structure.
Definition at line 249 of file res_speech.c.
References ast_set_flag, AST_SPEECH_SPOKE, AST_SPEECH_STATE_WAIT, and ast_speech::state.
Referenced by ast_speech_new(), handle_speechrecognize(), and speech_background().
00250 { 00251 int res = 0; 00252 00253 if (state == AST_SPEECH_STATE_WAIT) { 00254 /* The engine heard audio, so they spoke */ 00255 ast_set_flag(speech, AST_SPEECH_SPOKE); 00256 } 00257 00258 speech->state = state; 00259 00260 return res; 00261 }
int ast_speech_destroy | ( | struct ast_speech * | speech | ) |
Destroy a speech structure.
Definition at line 224 of file res_speech.c.
References ast_free, ast_mutex_destroy, ast_speech_results_free(), ast_speech_engine::destroy, ast_speech::engine, ast_speech::lock, ast_speech::processing_sound, and ast_speech::results.
Referenced by destroy_callback(), handle_speechdestroy(), launch_asyncagi(), run_agi(), and speech_create().
00225 { 00226 int res = 0; 00227 00228 /* Call our engine so we are destroyed properly */ 00229 speech->engine->destroy(speech); 00230 00231 /* Deinitialize the lock */ 00232 ast_mutex_destroy(&speech->lock); 00233 00234 /* If results exist on the speech structure, destroy them */ 00235 if (speech->results) 00236 ast_speech_results_free(speech->results); 00237 00238 /* If a processing sound is set - free the memory used by it */ 00239 if (speech->processing_sound) 00240 ast_free(speech->processing_sound); 00241 00242 /* Aloha we are done */ 00243 ast_free(speech); 00244 00245 return res; 00246 }
int ast_speech_dtmf | ( | struct ast_speech * | speech, | |
const char * | dtmf | |||
) |
Signal to the engine that DTMF was received.
Definition at line 155 of file res_speech.c.
References AST_SPEECH_STATE_READY, ast_speech_engine::dtmf, ast_speech::engine, and ast_speech::state.
Referenced by speech_background().
int ast_speech_grammar_activate | ( | struct ast_speech * | speech, | |
const char * | grammar_name | |||
) |
Activate a grammar on a speech structure.
Definition at line 67 of file res_speech.c.
References ast_speech_engine::activate, and ast_speech::engine.
Referenced by handle_speechactivategrammar(), and speech_activate().
int ast_speech_grammar_deactivate | ( | struct ast_speech * | speech, | |
const char * | grammar_name | |||
) |
Deactivate a grammar on a speech structure.
Definition at line 73 of file res_speech.c.
References ast_speech_engine::deactivate, and ast_speech::engine.
Referenced by handle_speechdeactivategrammar(), and speech_deactivate().
00074 { 00075 return (speech->engine->deactivate ? speech->engine->deactivate(speech, grammar_name) : -1); 00076 }
int ast_speech_grammar_load | ( | struct ast_speech * | speech, | |
const char * | grammar_name, | |||
const char * | grammar | |||
) |
Load a grammar on a speech structure (not globally).
Definition at line 79 of file res_speech.c.
References ast_speech::engine, and ast_speech_engine::load.
Referenced by handle_speechloadgrammar(), and speech_load().
int ast_speech_grammar_unload | ( | struct ast_speech * | speech, | |
const char * | grammar_name | |||
) |
Unload a grammar.
Definition at line 85 of file res_speech.c.
References ast_speech::engine, and ast_speech_engine::unload.
Referenced by handle_speechunloadgrammar(), and speech_unload().
struct ast_speech* ast_speech_new | ( | const char * | engine_name, | |
int | formats | |||
) | [read] |
Create a new speech structure.
Definition at line 176 of file res_speech.c.
References ast_best_codec(), ast_calloc, AST_FORMAT_SLINEAR, ast_free, ast_mutex_destroy, ast_mutex_init, ast_speech_change_state(), AST_SPEECH_STATE_NOT_READY, ast_speech_engine::create, ast_speech::engine, find_engine(), ast_speech::format, format, ast_speech_engine::formats, ast_speech::lock, and ast_speech::results.
Referenced by handle_speechcreate(), and speech_create().
00177 { 00178 struct ast_speech_engine *engine = NULL; 00179 struct ast_speech *new_speech = NULL; 00180 int format = AST_FORMAT_SLINEAR; 00181 00182 /* Try to find the speech recognition engine that was requested */ 00183 if (!(engine = find_engine(engine_name))) 00184 return NULL; 00185 00186 /* Before even allocating the memory below do some codec negotiation, we choose the best codec possible and fall back to signed linear if possible */ 00187 if ((format = (engine->formats & formats))) 00188 format = ast_best_codec(format); 00189 else if ((engine->formats & AST_FORMAT_SLINEAR)) 00190 format = AST_FORMAT_SLINEAR; 00191 else 00192 return NULL; 00193 00194 /* Allocate our own speech structure, and try to allocate a structure from the engine too */ 00195 if (!(new_speech = ast_calloc(1, sizeof(*new_speech)))) 00196 return NULL; 00197 00198 /* Initialize the lock */ 00199 ast_mutex_init(&new_speech->lock); 00200 00201 /* Make sure no results are present */ 00202 new_speech->results = NULL; 00203 00204 /* Copy over our engine pointer */ 00205 new_speech->engine = engine; 00206 00207 /* Can't forget the format audio is going to be in */ 00208 new_speech->format = format; 00209 00210 /* We are not ready to accept audio yet */ 00211 ast_speech_change_state(new_speech, AST_SPEECH_STATE_NOT_READY); 00212 00213 /* Pass ourselves to the engine so they can set us up some more and if they error out then do not create a structure */ 00214 if (engine->create(new_speech, format)) { 00215 ast_mutex_destroy(&new_speech->lock); 00216 ast_free(new_speech); 00217 new_speech = NULL; 00218 } 00219 00220 return new_speech; 00221 }
int ast_speech_register | ( | struct ast_speech_engine * | engine | ) |
Register a speech recognition engine.
Definition at line 272 of file res_speech.c.
References ast_log(), AST_RWLIST_INSERT_HEAD, AST_RWLIST_UNLOCK, AST_RWLIST_WRLOCK, ast_verb, ast_speech_engine::create, ast_speech_engine::destroy, find_engine(), LOG_WARNING, ast_speech_engine::name, and ast_speech_engine::write.
00273 { 00274 int res = 0; 00275 00276 /* Confirm the engine meets the minimum API requirements */ 00277 if (!engine->create || !engine->write || !engine->destroy) { 00278 ast_log(LOG_WARNING, "Speech recognition engine '%s' did not meet minimum API requirements.\n", engine->name); 00279 return -1; 00280 } 00281 00282 /* If an engine is already loaded with this name, error out */ 00283 if (find_engine(engine->name)) { 00284 ast_log(LOG_WARNING, "Speech recognition engine '%s' already exists.\n", engine->name); 00285 return -1; 00286 } 00287 00288 ast_verb(2, "Registered speech recognition engine '%s'\n", engine->name); 00289 00290 /* Add to the engine linked list and make default if needed */ 00291 AST_RWLIST_WRLOCK(&engines); 00292 AST_RWLIST_INSERT_HEAD(&engines, engine, list); 00293 if (!default_engine) { 00294 default_engine = engine; 00295 ast_verb(2, "Made '%s' the default speech recognition engine\n", engine->name); 00296 } 00297 AST_RWLIST_UNLOCK(&engines); 00298 00299 return res; 00300 }
int ast_speech_results_free | ( | struct ast_speech_result * | result | ) |
Free a set of results.
Definition at line 97 of file res_speech.c.
References ast_free, AST_LIST_NEXT, ast_speech_result::grammar, ast_speech_result::list, and ast_speech_result::text.
Referenced by ast_speech_destroy(), and ast_speech_start().
00098 { 00099 struct ast_speech_result *current_result = result, *prev_result = NULL; 00100 int res = 0; 00101 00102 while (current_result != NULL) { 00103 prev_result = current_result; 00104 /* Deallocate what we can */ 00105 if (current_result->text != NULL) { 00106 ast_free(current_result->text); 00107 current_result->text = NULL; 00108 } 00109 if (current_result->grammar != NULL) { 00110 ast_free(current_result->grammar); 00111 current_result->grammar = NULL; 00112 } 00113 /* Move on and then free ourselves */ 00114 current_result = AST_LIST_NEXT(current_result, list); 00115 ast_free(prev_result); 00116 prev_result = NULL; 00117 } 00118 00119 return res; 00120 }
struct ast_speech_result* ast_speech_results_get | ( | struct ast_speech * | speech | ) | [read] |
Get speech recognition results.
Definition at line 91 of file res_speech.c.
References ast_speech::engine, and ast_speech_engine::get.
Referenced by handle_speechrecognize(), and speech_background().
void ast_speech_start | ( | struct ast_speech * | speech | ) |
Indicate to the speech engine that audio is now going to start being written.
Definition at line 123 of file res_speech.c.
References ast_clear_flag, AST_SPEECH_HAVE_RESULTS, AST_SPEECH_QUIET, ast_speech_results_free(), AST_SPEECH_SPOKE, ast_speech::engine, ast_speech::results, and ast_speech_engine::start.
Referenced by handle_speechrecognize(), speech_background(), and speech_start().
00124 { 00125 00126 /* Clear any flags that may affect things */ 00127 ast_clear_flag(speech, AST_SPEECH_SPOKE); 00128 ast_clear_flag(speech, AST_SPEECH_QUIET); 00129 ast_clear_flag(speech, AST_SPEECH_HAVE_RESULTS); 00130 00131 /* If results are on the structure, free them since we are starting again */ 00132 if (speech->results) { 00133 ast_speech_results_free(speech->results); 00134 speech->results = NULL; 00135 } 00136 00137 /* If the engine needs to start stuff up, do it */ 00138 if (speech->engine->start) 00139 speech->engine->start(speech); 00140 00141 return; 00142 }
int ast_speech_unregister | ( | const char * | engine_name | ) |
Unregister a speech recognition engine.
Definition at line 303 of file res_speech.c.
References AST_RWLIST_FIRST, AST_RWLIST_REMOVE_CURRENT, AST_RWLIST_TRAVERSE_SAFE_BEGIN, AST_RWLIST_TRAVERSE_SAFE_END, AST_RWLIST_UNLOCK, AST_RWLIST_WRLOCK, ast_strlen_zero(), ast_verb, ast_speech_engine::list, and ast_speech_engine::name.
00304 { 00305 struct ast_speech_engine *engine = NULL; 00306 int res = -1; 00307 00308 if (ast_strlen_zero(engine_name)) 00309 return -1; 00310 00311 AST_RWLIST_WRLOCK(&engines); 00312 AST_RWLIST_TRAVERSE_SAFE_BEGIN(&engines, engine, list) { 00313 if (!strcasecmp(engine->name, engine_name)) { 00314 /* We have our engine... removed it */ 00315 AST_RWLIST_REMOVE_CURRENT(list); 00316 /* If this was the default engine, we need to pick a new one */ 00317 if (engine == default_engine) { 00318 default_engine = AST_RWLIST_FIRST(&engines); 00319 } 00320 ast_verb(2, "Unregistered speech recognition engine '%s'\n", engine_name); 00321 /* All went well */ 00322 res = 0; 00323 break; 00324 } 00325 } 00326 AST_RWLIST_TRAVERSE_SAFE_END; 00327 AST_RWLIST_UNLOCK(&engines); 00328 00329 return res; 00330 }
int ast_speech_write | ( | struct ast_speech * | speech, | |
void * | data, | |||
int | len | |||
) |
Write audio to the speech engine.
Definition at line 145 of file res_speech.c.
References AST_SPEECH_STATE_READY, ast_speech::engine, ast_speech::state, and ast_speech_engine::write.
Referenced by handle_speechrecognize(), and speech_background().
00146 { 00147 /* Make sure the speech engine is ready to accept audio */ 00148 if (speech->state != AST_SPEECH_STATE_READY) 00149 return -1; 00150 00151 return speech->engine->write(speech, data, len); 00152 }