Public Types
using	Initiator = SpeechRecognizerEngineInterface::Initiator

Public Member Functions
bool	holdToTalk ()

bool	tapToTalk ()

bool	startCapture (Initiator initiator, uint64_t keywordBegin=UNSPECIFIED_INDEX, uint64_t keywordEnd=UNSPECIFIED_INDEX, const std::string &keyword="")

bool	stopCapture ()

virtual bool	wakewordDetected (const std::string &wakeword)

virtual void	endOfSpeechDetected ()

Detailed Description

SpeechRecognizer should be extended to initiate voice interactions with Alexa.

SpeechRecognizer provides interfaces for

Initiating a dialog interaction with Alexa via press-and-hold, press-and-release, and voice-initiated user actions
Streaming audio input to the Engine
Enabling and disabling the wake word engine
Handling wake word and end of speech detection

Note: For observing Alexa dialog state transitions, see AlexaClient::dialogStateChanged().

Deprecated:: This platform interface is deprecated. Use the Alexa Auto Services Bridge (AASB) message broker to publish and subscribe to AASB messages instead.

See also: aace::core::MessageBroker

Member Typedef Documentation

◆ Initiator

using aace::alexa::SpeechRecognizer::Initiator = SpeechRecognizerEngineInterface::Initiator

Describes type of event that initiated the speech request

See also: aace::alexa::SpeechRecognizerEngineInterface::Initiator

Member Function Documentation

◆ holdToTalk()

bool aace::alexa::SpeechRecognizer::holdToTalk ( )

Notifies the Engine of a speech recognition event initiated by a press-and-hold action on the platform. The Engine will call aace::audio::AudioInput::startAudioInput() to notify the platform implementation when to start writing audio samples.

The platform implementation should call stopCapture() to terminate speech recognition on release of the press-and-hold action.

Returns: true if the Engine successfully initiated a recognize event, else false

◆ tapToTalk()

bool aace::alexa::SpeechRecognizer::tapToTalk ( )

Notifies the Engine of a speech recognition event initiated by a press-and-release action on the platform. The Engine will call aace::audio::AudioInput::startAudioInput() to notify the platform implementation when to start writing audio samples.

The Engine will terminate the recognize event initiated by the press-and-release action when end of speech is detected.

Returns: true if the Engine successfully started a recognize event, else false

◆ startCapture()

bool aace::alexa::SpeechRecognizer::startCapture	(	Initiator	initiator,
		uint64_t	keywordBegin = `UNSPECIFIED_INDEX`,
		uint64_t	keywordEnd = `UNSPECIFIED_INDEX`,
		const std::string &	keyword = `""`
	)

Notifies the Engine of a speech recognition event.

If the initiator type is HOLD_TO_TALK, then the platform implementation should call stopCapture() to terminate speech recognition on release of the press-and-hold action. Otherwise, the Engine will terminate the recognize event when end of speech is detected.

Parameters

[in]	initiator	The `Initiator` type for the speech recognition event
[in]	keywordBegin	The sample index where the keyword begins. This is required when the initiator type is `WAKEWORD`, otherwise should be set to `SpeechRecognizer::UNSPECIFIED_INDEX`.
[in]	keywordEnd	The sample index where the keyword ends. This is required when the initiator type is `WAKEWORD`, otherwise should be set to `SpeechRecognizer::UNSPECIFIED_INDEX`.
[in]	keyword	The keyword being recognized, e.g. "alexa". This is required when the initiator type is `WAKEWORD`, otherwise should be set to an empty string.

Returns: true if the Engine successfully started a recognize event, else false

◆ stopCapture()

bool aace::alexa::SpeechRecognizer::stopCapture ( )

Notifies the Engine to terminate the current recognize event. The Engine will call aace::audio::AudioInput::stopAudioInput() to notify the platform implementation when to stop writing audio samples.

Returns: true if the Engine successfully terminated the current recognize event, else false

◆ wakewordDetected()

bool aace::alexa::SpeechRecognizer::wakewordDetected ( const std::string & wakeword )

virtual

Notifies the platform implementation when a wake word is detected

Parameters

[in] wakeword The wake word that was detected

Returns: true if the Engine should initiate a recognize event, false if the Engine should ignore the invocation

◆ endOfSpeechDetected()

void aace::alexa::SpeechRecognizer::endOfSpeechDetected ( )