sound trigger: add sound model and key phrase IDs.
Bug: 12378680. Change-Id: I93b1916151108ccf6eb625b0078e6ed574720b76
This commit is contained in:
parent
019b8ece22
commit
db37858a17
|
@ -92,6 +92,7 @@ typedef int sound_model_handle_t;
|
|||
*/
|
||||
struct sound_trigger_sound_model {
|
||||
sound_trigger_sound_model_type_t type; /* model type. e.g. SOUND_MODEL_TYPE_KEYPHRASE */
|
||||
sound_trigger_uuid_t uuid; /* unique sound model ID. */
|
||||
unsigned int data_size; /* size of opaque model data */
|
||||
unsigned int data_offset; /* offset of opaque data start from head of struct
|
||||
(e.g sizeof struct sound_trigger_sound_model) */
|
||||
|
@ -99,6 +100,7 @@ struct sound_trigger_sound_model {
|
|||
|
||||
/* key phrase descriptor */
|
||||
struct sound_trigger_phrase {
|
||||
unsigned int id; /* keyphrase ID */
|
||||
unsigned int recognition_mode; /* recognition modes supported by this key phrase */
|
||||
unsigned int num_users; /* number of users in the key phrase */
|
||||
char locale[SOUND_TRIGGER_MAX_LOCALE_LEN]; /* locale - JAVA Locale style (e.g. en_US) */
|
||||
|
|
Loading…
Reference in New Issue