Declared in BZMediaParams.h

BZSRTaskType

This constant enumerator defines the Speech Recognition Task Type (SRTT) used to perform speech recognition. For Apple Speech only SRTT_LM task type is supported.

Definition

typedef NS_ENUM(NSInteger, BZSRTaskType ) {
   SRTT_UNKNOWN = -1,
   SRTT_JSGF = 0,
   SRTT_JSGF_STRING = 1,
   SRTT_FSG = 2,
   SRTT_LM = 3,
   SRTT_KEYPHRASE = 4,
   SRTT_KWS = 5,
   SRTT_ALLPHONE = 6,
};

Constants

SRTT_UNKNOWN

Speech Recognition Task Type is unknown. This constant should not be used.

Declared In BZMediaParams.h.

SRTT_JSGF

Speech Recognition Task Type is JSpeech Grammar Format JSGF text contained in a file that can be accessed by a speech recognizer. Also referred to as a speech recognition grammar, JSGF defines the set of possible words and sentences that can recognized while adhering to rules. For more information please refer to CMUSphinx Grammar.

Declared In BZMediaParams.h.

SRTT_JSGF_STRING

Not supported by Apple Speech.

Speech Recognition Task Type is JSpeech Grammar Format JSGF text contained in a text string that can be passed to a speech recognizer. Also referred to as a speech recognition grammar, JSGF defines the set of possible words and sentences that can recognized while adhering to rules. For more information please refer to CMUSphinx Grammar.

Declared In BZMediaParams.h.

SRTT_FSG

Not supported by Apple Speech.

Speech Recognition Task Type is Finite State Grammar (FSG). Usually, FSG task type descriptions can be written by hand in the JSGF format instead of being used directly. JSGF grammars can be converted to FSG grammars using tools available from SphinxBase.

Declared In BZMediaParams.h.

SRTT_LM

Supported by both Apple Speech and PocketSphinx.

Speech Recognition Task Type is Language Model (LM) contained in a file that can be accessed by a speech recognizer. Also referred to as a N-Gram model, this model is trained from corpora of text.

For PocketSphinx: BZMedia Framework includes default english and spanish language models for SRE_POCKET_SPHINX speech recognition engine. Additional PockeSphinx Language Models can be built by following the instructions detailed in CMUSphinx Language Model documentation. Other pre-built language models and supporting different languages can be downloaded from SphinxBase and other areas on the web.

For Apple Speech: Apple supports speech recognition for many languages using Apple hosted speech Language Models. This is the only Speech Recognition Task Type supported by Apple Speech.

Declared In BZMediaParams.h.

SRTT_KEYPHRASE

Not supported by Apple Speech.

Speech Recognition Task Type is Key Phrase which is one phrase that the speech recognizer can only recognize. This phrase is passed to the speech recognizer from an application. For more information, please review source code provided by PocketSphinx and SphinxBase.

Declared In BZMediaParams.h.

SRTT_KWS

Not supported by Apple Speech.

Speech Recognition Task Type is Key Word Spotting (KWS) which is the speech recognition of multiple words and phrases while allowing the speech to be recognized to contain extra speech and sounds that surround the words and phrases to be recognized. For more information, please review source code provided by PocketSphinx and SphinxBase.

Declared In BZMediaParams.h.

SRTT_ALLPHONE

Not supported by Apple Speech.

Speech Recognition Task Type is All Phone which is the speech recognition of the speech phonemes found in the acoustic model that’s used for speech recognition. For more information, please review source code provided by PocketSphinx and SphinxBase.

Declared In BZMediaParams.h.

Declared In

BZMediaParams.h