Inherits from NSObject
Declared in ISSpeechRecognition.h

Overview

The interface for doing speech recognition in the SDK.

Tasks

Getting and Setting the Delegate

  •   delegate

    The delegate of a speech recognition object.

         property

Configuration Properties

  •   locale

    Sets the locale to use for speech recognition.

         property
  •   model

    Allows you to set a custom language model for speech recognition.

         property
  •   freeformType

    The type of model to use when trascribing audio. Defaults to ISFreeFormTypeDictation.

         property
  •   silenceDetectionEnabled

    Whether silence detection should be used to automatically detect when someone’s done talking.

         property

Detecting Audio Input

  • + audioInputAvailable

    Returns whether audio input is available for speech recognition. You can check this before creating an instance of ISSpeechRecognition, as well as to dynamically update your UI with what you can do.

Aliases and Commands

Listening and Recognizing

Properties

delegate

The delegate of a speech recognition object.

@property (nonatomic, unsafe_unretained) id<ISSpeechRecognitionDelegate> delegate

Discussion

The delegate must adopt the ISSpeechRecognitionDelegate protocol.

Declared In

ISSpeechRecognition.h

freeformType

The type of model to use when trascribing audio. Defaults to ISFreeFormTypeDictation.

@property (nonatomic, assign) ISFreeFormType freeformType

Declared In

ISSpeechRecognition.h

locale

Sets the locale to use for speech recognition.

@property (nonatomic, copy) NSString *locale

Discussion

Most of the time, the value passed is a ISO country code. To get our supported ISOs, consult “Freeform Dictation Languages” under “Speech Recognition Settings” when viewing details about a specific key.

Declared In

ISSpeechRecognition.h

model

Allows you to set a custom language model for speech recognition.

@property (nonatomic, copy) NSString *model

Declared In

ISSpeechRecognition.h

silenceDetectionEnabled

Whether silence detection should be used to automatically detect when someone’s done talking.

@property (nonatomic, assign) BOOL silenceDetectionEnabled

Declared In

ISSpeechRecognition.h

Class Methods

audioInputAvailable

Returns whether audio input is available for speech recognition. You can check this before creating an instance of ISSpeechRecognition, as well as to dynamically update your UI with what you can do.

+ (BOOL)audioInputAvailable

Return Value

Returns whether audio input is available.

Declared In

ISSpeechRecognition.h

Instance Methods

addAlias:forItems:

Adds a list of items as an alias.

- (void)addAlias:(NSString *)alias forItems:(NSArray *)items

Parameters

alias

The string to use for the alias key.

items

The array of items to be substituted for the alias.

Discussion

Think of using an alias list as a way of doing a regular expression. For example, if you want to use a regular expression to match “call joe”, “call charlie”, or “call ben”, then it would be call (joe|charlie|ben). Similarly, to do that with an alias list,

[speechRecognitionInstance addAlias:@"PEOPLE" forItems:[NSArray arrayWithObjects:@"joe", @"charlie", @"ben", nil]];
[speechRecognitionInstance addCommand:@"call %PEOPLE%"];

Declared In

ISSpeechRecognition.h

addCommand:

Adds a command.

- (void)addCommand:(NSString *)command

Parameters

command

The command to be added.

Discussion

If you want to reference an alias list, the format is %ALIAS_LIST_NAME%. Replace ALIAS_LIST_NAME with the actual name of your alias list.

Declared In

ISSpeechRecognition.h

addCommands:

Add multiple commands from an array.

- (void)addCommands:(NSArray *)commands

Parameters

commands

The array of commands to be added.

Declared In

ISSpeechRecognition.h

cancel

Cancels an in progress speech recognition action.

- (void)cancel

Discussion

If, for some reason, you need to cancel an in progress speech recognition action, call this method. It’s also a good idea to provide feedback to the user as to why you cancelled it.

Declared In

ISSpeechRecognition.h

finishListenAndStartRecognize

If you’re running an untimed listen, or if you want to cut a timed listen short, call this method to tell the SDK to stop listening for audio, and finish up transcribing.

- (void)finishListenAndStartRecognize

Declared In

ISSpeechRecognition.h

listen:

Start an untimed listen.

- (BOOL)listen:(NSError **)err

Parameters

err

An NSError pointer to get an error object out of the method if something goes wrong.

Return Value

Returns whatever speech synthesis successfully started. If this returns NO, check the error for details on what went wrong.

Discussion

An untimed listen means that the SDK will start listening, and will not stop unless you tell it to by calling –[ISSpeechRecognition finishListenAndStartRecognize], or until silence detection kicks in, if you have that enabled.

If you’re using a command or alias list, use -listenAndRecognizeWithTimeout:error: instead. This will ensure that speech recognition will only last as long as is necessary, thus saving the user’s battery life and data plan, ensuring that you get a result back, and providing an better overall experience.

See Also

Declared In

ISSpeechRecognition.h

listenAndRecognizeWithTimeout:error:

Starts a timed listen. After a set timeout, the SDK will stop listening for audio and will start to transcribe it.

- (BOOL)listenAndRecognizeWithTimeout:(NSTimeInterval)timeout error:(NSError **)err

Parameters

timeout

The amount of time, in seconds, for the timed listen to last for.

err

An NSError pointer to get an error object out of the method if something goes wrong.

Return Value

Returns whatever speech synthesis successfully started. If this returns NO, check the error for details on what went wrong.

Discussion

Useful when using command lists to ensure that the user doesn’t talk longer than necessary.

Declared In

ISSpeechRecognition.h

listenAndRecognizeWithTimeout:handler:

Starts a timed listen. After a set timeout, the SDK will stop listening for audio and will start to transcribe it.

- (void)listenAndRecognizeWithTimeout:(NSTimeInterval)timeout handler:(ISSpeechRecognitionHandler)handler

Parameters

timeout

The amount of time, in seconds, for the timed listen to last for.

handler

An ISSpeechRecognitionHandler block that will be executed on the main thread when speech recognition completes, or when an error occurs.

Discussion

Useful when using command lists to ensure that the user doesn’t talk longer than necessary.

Declared In

ISSpeechRecognition.h

listenWithHandler:

Start an untimed listen.

- (void)listenWithHandler:(ISSpeechRecognitionHandler)handler

Parameters

handler

An ISSpeechRecognitionHandler block that will be executed on the main thread when speech recognition completes, or when an error occurs.

Discussion

An untimed listen means that the SDK will start listening, and will not stop unless you tell it to by calling –[ISSpeechRecognition finishListenAndStartRecognize], or until silence detection kicks in, if you have that enabled.

If you’re using a command or alias list, use -listenAndRecognizeWithTimeout:handler: instead. This will ensure that speech recognition will only last as long as is necessary, thus saving the user’s battery life and data plan, ensuring that you get a result back, and providing an better overall experience.

See Also

Declared In

ISSpeechRecognition.h

resetCommandsAndAliases

Clears any command or alias lists on this speech recognition object.

- (void)resetCommandsAndAliases

Declared In

ISSpeechRecognition.h