1/* 2 * Copyright (C) 2014 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17#ifndef ANDROID_SOUND_TRIGGER_H 18#define ANDROID_SOUND_TRIGGER_H 19 20#include <stdbool.h> 21#include <system/audio.h> 22 23#define SOUND_TRIGGER_MAX_STRING_LEN 64 /* max length of strings in properties or 24 descriptor structs */ 25#define SOUND_TRIGGER_MAX_LOCALE_LEN 6 /* max length of locale string. e.g en_US */ 26#define SOUND_TRIGGER_MAX_USERS 10 /* max number of concurrent users */ 27#define SOUND_TRIGGER_MAX_PHRASES 10 /* max number of concurrent phrases */ 28 29typedef enum { 30 SOUND_TRIGGER_STATE_NO_INIT = -1, /* The sound trigger service is not initialized */ 31 SOUND_TRIGGER_STATE_ENABLED = 0, /* The sound trigger service is enabled */ 32 SOUND_TRIGGER_STATE_DISABLED = 1 /* The sound trigger service is disabled */ 33} sound_trigger_service_state_t; 34 35#define RECOGNITION_MODE_VOICE_TRIGGER 0x1 /* simple voice trigger */ 36#define RECOGNITION_MODE_USER_IDENTIFICATION 0x2 /* trigger only if one user in model identified */ 37#define RECOGNITION_MODE_USER_AUTHENTICATION 0x4 /* trigger only if one user in mode 38 authenticated */ 39#define RECOGNITION_STATUS_SUCCESS 0 40#define RECOGNITION_STATUS_ABORT 1 41#define RECOGNITION_STATUS_FAILURE 2 42 43#define SOUND_MODEL_STATUS_UPDATED 0 44 45typedef enum { 46 SOUND_MODEL_TYPE_UNKNOWN = -1, /* use for unspecified sound model type */ 47 SOUND_MODEL_TYPE_KEYPHRASE = 0 /* use for key phrase sound models */ 48} sound_trigger_sound_model_type_t; 49 50typedef struct sound_trigger_uuid_s { 51 unsigned int timeLow; 52 unsigned short timeMid; 53 unsigned short timeHiAndVersion; 54 unsigned short clockSeq; 55 unsigned char node[6]; 56} sound_trigger_uuid_t; 57 58/* 59 * sound trigger implementation descriptor read by the framework via get_properties(). 60 * Used by SoundTrigger service to report to applications and manage concurrency and policy. 61 */ 62struct sound_trigger_properties { 63 char implementor[SOUND_TRIGGER_MAX_STRING_LEN]; /* implementor name */ 64 char description[SOUND_TRIGGER_MAX_STRING_LEN]; /* implementation description */ 65 unsigned int version; /* implementation version */ 66 sound_trigger_uuid_t uuid; /* unique implementation ID. 67 Must change with version each version */ 68 unsigned int max_sound_models; /* maximum number of concurrent sound models 69 loaded */ 70 unsigned int max_key_phrases; /* maximum number of key phrases */ 71 unsigned int max_users; /* maximum number of concurrent users detected */ 72 unsigned int recognition_modes; /* all supported modes. 73 e.g RECOGNITION_MODE_VOICE_TRIGGER */ 74 bool capture_transition; /* supports seamless transition from detection 75 to capture */ 76 unsigned int max_buffer_ms; /* maximum buffering capacity in ms if 77 capture_transition is true*/ 78 bool concurrent_capture; /* supports capture by other use cases while 79 detection is active */ 80 bool trigger_in_event; /* returns the trigger capture in event */ 81 unsigned int power_consumption_mw; /* Rated power consumption when detection is active 82 with TDB silence/sound/speech ratio */ 83}; 84 85typedef int sound_trigger_module_handle_t; 86 87struct sound_trigger_module_descriptor { 88 sound_trigger_module_handle_t handle; 89 struct sound_trigger_properties properties; 90}; 91 92typedef int sound_model_handle_t; 93 94/* 95 * Generic sound model descriptor. This struct is the header of a larger block passed to 96 * load_sound_model() and containing the binary data of the sound model. 97 * Proprietary representation of users in binary data must match information indicated 98 * by users field 99 */ 100struct sound_trigger_sound_model { 101 sound_trigger_sound_model_type_t type; /* model type. e.g. SOUND_MODEL_TYPE_KEYPHRASE */ 102 sound_trigger_uuid_t uuid; /* unique sound model ID. */ 103 sound_trigger_uuid_t vendor_uuid; /* unique vendor ID. Identifies the engine the 104 sound model was build for */ 105 unsigned int data_size; /* size of opaque model data */ 106 unsigned int data_offset; /* offset of opaque data start from head of struct 107 (e.g sizeof struct sound_trigger_sound_model) */ 108}; 109 110/* key phrase descriptor */ 111struct sound_trigger_phrase { 112 unsigned int id; /* keyphrase ID */ 113 unsigned int recognition_mode; /* recognition modes supported by this key phrase */ 114 unsigned int num_users; /* number of users in the key phrase */ 115 unsigned int users[SOUND_TRIGGER_MAX_USERS]; /* users ids: (not uid_t but sound trigger 116 specific IDs */ 117 char locale[SOUND_TRIGGER_MAX_LOCALE_LEN]; /* locale - JAVA Locale style (e.g. en_US) */ 118 char text[SOUND_TRIGGER_MAX_STRING_LEN]; /* phrase text in UTF-8 format. */ 119}; 120 121/* 122 * Specialized sound model for key phrase detection. 123 * Proprietary representation of key phrases in binary data must match information indicated 124 * by phrases field 125 */ 126struct sound_trigger_phrase_sound_model { 127 struct sound_trigger_sound_model common; 128 unsigned int num_phrases; /* number of key phrases in model */ 129 struct sound_trigger_phrase phrases[SOUND_TRIGGER_MAX_PHRASES]; 130}; 131 132 133/* 134 * Generic recognition event sent via recognition callback 135 */ 136struct sound_trigger_recognition_event { 137 int status; /* recognition status e.g. 138 RECOGNITION_STATUS_SUCCESS */ 139 sound_trigger_sound_model_type_t type; /* event type, same as sound model type. 140 e.g. SOUND_MODEL_TYPE_KEYPHRASE */ 141 sound_model_handle_t model; /* loaded sound model that triggered the 142 event */ 143 bool capture_available; /* it is possible to capture audio from this 144 utterance buffered by the 145 implementation */ 146 int capture_session; /* audio session ID. framework use */ 147 int capture_delay_ms; /* delay in ms between end of model 148 detection and start of audio available 149 for capture. A negative value is possible 150 (e.g. if key phrase is also available for 151 capture */ 152 int capture_preamble_ms; /* duration in ms of audio captured 153 before the start of the trigger. 154 0 if none. */ 155 bool trigger_in_data; /* the opaque data is the capture of 156 the trigger sound */ 157 audio_config_t audio_config; /* audio format of either the trigger in 158 event data or to use for capture of the 159 rest of the utterance */ 160 unsigned int data_size; /* size of opaque event data */ 161 unsigned int data_offset; /* offset of opaque data start from start of 162 this struct (e.g sizeof struct 163 sound_trigger_phrase_recognition_event) */ 164}; 165 166/* 167 * Confidence level for each user in struct sound_trigger_phrase_recognition_extra 168 */ 169struct sound_trigger_confidence_level { 170 unsigned int user_id; /* user ID */ 171 unsigned int level; /* confidence level in percent (0 - 100). 172 - min level for recognition configuration 173 - detected level for recognition event */ 174}; 175 176/* 177 * Specialized recognition event for key phrase detection 178 */ 179struct sound_trigger_phrase_recognition_extra { 180 unsigned int id; /* keyphrase ID */ 181 unsigned int recognition_modes; /* recognition modes used for this keyphrase */ 182 unsigned int confidence_level; /* confidence level for mode RECOGNITION_MODE_VOICE_TRIGGER */ 183 unsigned int num_levels; /* number of user confidence levels */ 184 struct sound_trigger_confidence_level levels[SOUND_TRIGGER_MAX_USERS]; 185}; 186 187struct sound_trigger_phrase_recognition_event { 188 struct sound_trigger_recognition_event common; 189 unsigned int num_phrases; 190 struct sound_trigger_phrase_recognition_extra phrase_extras[SOUND_TRIGGER_MAX_PHRASES]; 191}; 192 193/* 194 * configuration for sound trigger capture session passed to start_recognition() 195 */ 196struct sound_trigger_recognition_config { 197 audio_io_handle_t capture_handle; /* IO handle that will be used for capture. 198 N/A if capture_requested is false */ 199 audio_devices_t capture_device; /* input device requested for detection capture */ 200 bool capture_requested; /* capture and buffer audio for this recognition 201 instance */ 202 unsigned int num_phrases; /* number of key phrases recognition extras */ 203 struct sound_trigger_phrase_recognition_extra phrases[SOUND_TRIGGER_MAX_PHRASES]; 204 /* configuration for each key phrase */ 205 unsigned int data_size; /* size of opaque capture configuration data */ 206 unsigned int data_offset; /* offset of opaque data start from start of this struct 207 (e.g sizeof struct sound_trigger_recognition_config) */ 208}; 209 210/* 211 * Event sent via load sound model callback 212 */ 213struct sound_trigger_model_event { 214 int status; /* sound model status e.g. SOUND_MODEL_STATUS_UPDATED */ 215 sound_model_handle_t model; /* loaded sound model that triggered the event */ 216 unsigned int data_size; /* size of event data if any. Size of updated sound model if 217 status is SOUND_MODEL_STATUS_UPDATED */ 218 unsigned int data_offset; /* offset of data start from start of this struct 219 (e.g sizeof struct sound_trigger_model_event) */ 220}; 221 222 223#endif // ANDROID_SOUND_TRIGGER_H 224