io.spokestack.spokestack.android

Class AndroidSpeechRecognizer

java.lang.Object

io.spokestack.spokestack.android.AndroidSpeechRecognizer

All Implemented Interfaces:

SpeechProcessor, AutoCloseable

public class AndroidSpeechRecognizer
extends Object
implements SpeechProcessor

Speech recognition using built-in Android APIs.

This component uses the built-in Android SpeechRecognizer to process user speech.

As part of normal operation, SpeechRecognizer plays system sounds both when it starts and stops actively listening to the user, just like the built-in Google Assistant. This behavior is not optional; it can be suppressed by having the AudioManager mute the music stream, but it muting and restoring the volume of that stream at exactly the right times is error-prone, so such behavior has been omitted from this component.

Note that this component requires an Android Context to be attached to the pipeline that has created it. If the pipeline is meant to persist across different Activitys, the Context used must either be the application context, or it must be re-set on the pipeline's SpeechContext object when the Activity context changes.

Implementation of SpeechRecognizer is left up to devices, and even though the API exists, an actual recognizer may not be present on all devices. If using this component, it's a good idea to call SpeechRecognizer.isRecognitionAvailable() before adding it to the pipeline to determine whether it will be viable on the current device.

In addition, testing has shown that some older devices may return true for the preceding call but have outdated implementations that consistently throw errors. For this reason, it's a good idea to have an OnSpeechEventListener set up to detect SpeechRecognizerErrors and have an appropriate fallback strategy in place.

This pipeline component supports the following configuration property, though it should be left at its default setting in most circumstances:

• wake-active-min (integer): the minimum length of time, in milliseconds, that the recognizer will wait for speech before timing out.

Constructor Summary

Constructors

Constructor and Description
AndroidSpeechRecognizer(SpeechConfig speechConfig) Initializes a new recognizer.

Method Summary

Modifier and Type	Method and Description
void	close()
void	process(SpeechContext context, ByteBuffer frame) processes the current speech frame.
void	reset() resets all state internal to the stage.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AndroidSpeechRecognizer

public AndroidSpeechRecognizer(SpeechConfig speechConfig)

Initializes a new recognizer.

Parameters:

speechConfig - Spokestack pipeline configuration

Method Detail

process

public void process(SpeechContext context,
                    ByteBuffer frame)

Description copied from interface: SpeechProcessor

processes the current speech frame.

Specified by:

process in interface SpeechProcessor

Parameters:

context - the current speech context

frame - the received audio frame

reset

public void reset()

Description copied from interface: SpeechProcessor

resets all state internal to the stage.

Specified by:

reset in interface SpeechProcessor

close

public void close()

Specified by:

close in interface AutoCloseable