AVS Device SDK Release Notes
The following release notes contain details about the Alexa Voice Service (AVS) Device SDK, including feature enhancements, updates, and resolved issues.
2020
Version 1.22.0
Release date
December 8, 2020
Enhancements
- Added support for the AudioPlayer 1.5 interface.
- Added support for the System 2.1 interface. This allows you to report network related information.
- Added the
LibcurlSetCurlOptionsCallbackInterface
API toLibcurlHTTPConnectionFactory
. This allows you to accept custom curl options when establishing a connection. - Added the following components to the Manufactory subsystem.
- DoNotDisturb
- Alerts
- DefaultDeviceSettingsManager
- SystemTimeZone
- Added two new functions to
ChannelVolumeInterface
. These two new functions areadjustUnduckedVolume()
andgetId()
. UseChannelVolumeInterface::getId()
to dedupe ChannelVolumeInterfaces. - Improved reconnection behavior.
MessageRouter
now has grace period before it skips notifying observers of disconnected states.
Bug fixes
- Fixed an audio bug that could unexpectedly decrease the volume. This fix added an ID3TagsRemover to remove ID3 tags from Media Streams.
- Fixed a race condition in
CapabilitiesDelegate
that could send unnecessary Discovery events from the device. - Fixed an
SQLiteBluetoothStorage
bug that prevented it from checking for tables on initialization. - Fixed an
SQLiteNotificationsStorage
bug that prevented it from recovering when tables are missing. - Fixed a
DialogUXStateAggregator
bug that caused incorrect states.
New configurations
- The Default Client no longer closes
DeviceSettingStorageInterface
databases. Make sure your implementations ofDeviceSettingStorageInterface
closes theDeviceSettingStorageInterface
database in its own destructor.
Known issues
- On some compilers and platforms, the SampleApp might not run and provides the following manufactory-related error:
Recipes are not equivalent
. For more details about how to temporarily patch this problem, see issue 1824 on GitHub. This issue will be addressed in an upcoming release. - Build errors can occur on the Raspberry Pi due to incorrect linking of the atomic library. A suggested workaround is to add the following
set(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -latomic")
to the top most CMake file. - The WebVTT dependency required for captions isn't supported for Windows/Android.
- Exiting from the setting option takes you back to the Options Menu directly. It doesn't provide a message to indicate that you're back in the main menu.
- Disabled all failing Unit Tests and Audio Input Processor (AIP) Unit tests on Windows.
- AudioInputProcessor unit tests don't build on Windows when with the
-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter.v - Music playback history isn't displayed in the Alexa app for certain account and device types.
- The Gnu Compiler Collection (GCC) 8+
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this problem by upgrading to Android Debug Bridge (ADB) version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned though local Text To Speech (TTS).
- ACL encounters issues if it receives audio attachments but doesn't consume them.
- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- Interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
- Disabled
ServerDisconnectIntegratonTest
tests until they reflect the new service behavior. - The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently. - The Raspberry Pi sample app might crash after prolonged use due to conflicts with PortAudio dependencies.
- The sample app might crash if it's using libCurl 7.69.
Version 1.21.0
Release date
October 26, 2020
Enhancements
- Added support for Smart Home for AVS Semantics. For more details about SDK implementation, see Smart Home Endpoints.
- Added the following components to the Manufactory subsystem.
- AVSConnectionManager
- ApplicationAudioPipelineFactoryInterface (three full implementations – Gstreamer, Android, Custom – and a stub)
- Alexa Capability Agent
- CaptionManager
- CertifiedSender
- InternetConnectionMonitor
- Shutdown/StartupNotifier
- SpeakerManager
- AudioPlayer
- ExternalMediaPlayer
- FocusManager (Audio)
- Added support for User Perceived Latency Metrics. You must enable metrics to use this feature.
- Moved two Capability Agents to a different location, resulting in a change in name spaces and directory structures
- ExternalMediaPlayer – Moved from the CapabilityAgents/ExternalMediaPlayer subdirectory to the capabilities/ExternalMediaPlayer subdirectory.
- Equalizer – Moved from the CapabilityAgents/Equalizer subdirectory to the capabilities/Equalizer subdirectory.
- You can now initialize the
AlexaClientSDKInit
with theInitializationParameters
object. This object supports two additional objects:- TimerDelegateFactoryInterface
- PowerResourceManagerInterface.
- Introduced
TimerDelegate
abstraction. You can now inject your own timer logic and call it with existing AVSCommon/Utils/Timing/Timer.h objects. - Updated
DevicePropertyAggregator
to support the newRegistrationStatus
property. - Pulled in upstream RapidJSON patches
Bug fixes
- Android builds now support arm64 and x86_64 devices.
- Fixed an Android issue that caused false positive error logs when using the microphone.
- Fixed an audio injection issue that caused problems with multi-turn interactions when built with a disabled wake word.
- Fixed a sample app issue that caused the range controller menu to loop upon receiving invalid input.
- Fixed a sample app issue that caused a GuiRenderer crash when using certain directives.
- Add an optional
onSeekActivity
method to theAudioPlayerObserverInterface
so the AudioPlayer Capability Agent can notify observers when local seeks start and complete. - Added a new build flag
-DTARGET_SUFFIX="SomeString"
to suffix output library names, in case of naming conflicts. - Fixed an AudioPlayer issue that prevented the interface from transitioning to an IDLE state when it doesn't have any content to play.
- Decreased latency when playing short speak directives. For example "ok" or "hi".
- Fixed a music issue that prevented music playback from resuming after disabling ducking.
- Fixed a Bluetooth issue that caused incorrect information in the logs.
- Fixed a
DialogUXStateAggregator
bug that caused incorrect behavior when handling device states.
New configurations
- Added the
TARGET_RPATH
CMake parameter to specify a RPATH during build. - Added the
LPM
CMake parameter to enable or disable Low Power Mode. - Added the
DISABLE_DUCKING
CMake parameter to enable or disable ducking.
Known issues
- Build errors can occur on the Raspberry Pi due to incorrect linking of the atomic library. A suggested workaround is to add the following
set(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -latomic")
to the top most CMake file. - The WebVTT dependency required for captions isn't supported for Windows/Android.
- Exiting from the setting option takes you back to the Options Menu directly. It doesn't provide a message to indicate that you're back in the main menu.
- Disabled all failing Unit Tests and Audio Input Processor (AIP) Unit tests on Windows.
- AudioInputProcessor unit tests don't build on Windows when with the
-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter. - Music playback history isn't displayed in the Alexa app for certain account and device types.
- The Gnu Compiler Collection 8+ (GCC 8+)
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this problem by upgrading to Android Debug Bridge (ADB) version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned though local Text To Speech (TTS).
- ACL encounters issues if it receives audio attachments but doesn't consume them.
- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- Interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
- Disabled
ServerDisconnectIntegratonTest
tests until they reflect the new service behavior. - The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently. - The Raspberry Pi sample app might crash after prolonged use due to conflicts with PortAudio dependencies.
- The sample app might crash if it's using libCurl 7.69.
Version 1.20.1
Release date
August 6, 2020
Bug fixes
- The SDK component name changed from
com.amazon.sdk
tocom.amazon.alexa.deviceSDK
. This name gets reported through the Alexa.SoftwareComponentReporter interface. - Reverted a change that caused CertifiedSender to load messages from the database on initialization.
- Fixed a compilation error that occurred on Catalina macOS.
Version 1.20.0
Release date
June 22, 2020
Enhancements
- Added the Software Component Reporter Capability Agent (CA). This CA allows a device to report device software information to the Alexa Voice Service (AVS) using the Alexa.SoftwareComponentReporter interface.
- Added support for the SpeechSynthesizer 1.4 interface.
- Added support for dynamic endpoints. You can now add, modify, and delete endpoints at any point after the DefaultClient connects to AVS.
- Added support for custom media players. You can now swap out the default media player that's provided with the AVS Device SDK.
- Added the manufactory subsystem. This subsystem provides a more robust solution to manage and configure components within an Alexa app.
- Moved five Capability Agents to a different location, resulting in a change in name spaces and directory structures.
- Alerts – Moved from the CapabilityAgents/Alerts subdirectory to the capabilities/Alerts subdirectory.
- AudioPlayer – Moved from the CapabilityAgents/AudioPlayer subdirectory to the capabilities/AudioPlayer subdirectory.
- Bluetooth – Moved from the CapabilityAgents/Bluetooth subdirectory to the capabilities/Bluetooth subdirectory.
- Multi Room Music (MRM) – Moved from the CapabilityAgents/MRM subdirectory to the capabilities/MultiRoomMusic subdirectory.
- Notifications – Moved from the CapabilityAgents/Notifications subdirectory to the capabilities/Notifications subdirectory.
- Added the ability to update Alert tones with the SetAlert directive.
- Added support for local playback control with the new LocalPlaybackHandlerInterface.
- Added support for the SpeechSynthesizer 1.4 interface. This interface supports a feature that's in private beta and available in a future release.
New configurations
- Added the
-DACSDK_USE_RTTI
CMake parameter to toggle the use of Run Time Type Information (RTTI) in the SDK. This parameter defaults toOFF
, except for WIN32 builds, where it must beON
. - Added the
-DUSE_CCACHE
CMake parameter to build the SDK with ccache, which can help speed up compile time. This parameter defaults toOFF
.
DRAPIDJSON_MEM_OPTIMIZATION = ON
. This RapidJSON memory optimization configuration is now enabled on all builds by default. Double check that your code can change the RapidJSON memory allocator from MemoryPoolAllocator
to CrtAllocator
to avoid compilation errors.Bug fixes
- Fixed a Playlist Parser issue that caused a playlist URL to incorrectly resolve if it contained parameters and a relative path.
- Fixed a Gstreamer MediaPlayer implementation issue that resulted in a playback problem due to
seekOverwrittenData
errors. - Fixed an issue with
make install
where it didn't include the InterruptModelConfiguration.h file. - Fixed a Windows MSYS2 issue that caused a build to fail due to a Gnu Compiler Collection (GCC) linker timeout. An additional dependency – mingw-w64-x86_64-lld – is now required to use the SDK. Add this new dependency to MSYS2 if you are having trouble building the SDK.
- Fixed a timer issue that could cause a lockup or crash.
- Fixed an issue that incorrectly preserved the
dialogRequestId
across multi-turn interactions instead of generating a new one. - Fixed a buffer overflow in the
FFMpegInputBuffer
. - Added the missing HTTP Success Code 201 to the set of recognized HTTP response codes in the AVSCommon/Utils folder.
- Fixed a bug that caused a compile error if you enabled a single endpoint controller when building the SDK.
- Fixed an issue that prevents apps from adding or removing
ContextManager
observers with the DefaultClient.
Known issues
- Build errors can occur on the Raspberry Pi due to incorrect linking of the atomic library. A suggested workaround is to add the following
set(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -latomic")
to the top most CMake file. - The WebVTT dependency required for captions isn't supported for Windows/Android.
- Exiting from the setting option takes you back to the Options Menu directly. It doesn't provide a message to indicate that you're back in the main menu.
- Disabled all failing Unit Tests and Audio Input Processor (AIP) Unit tests on Windows.
- AudioInputProcessor unit tests don't build on Windows when with the
-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter. - Music playback history isn't displayed in the Alexa app for certain account and device types.
- The Gnu Compiler Collection 8+ (GCC 8+)
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this problem by upgrading to Android Debug Bridge (ADB) version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned though local Text To Speech (TTS).
- ACL encounters issues if it receives audio attachments but doesn't consume them.
- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- Interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
- Disabled
ServerDisconnectIntegratonTest
tests until they reflect the new service behavior. - The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently. - The Raspberry Pi sample app might crash after prolonged use due to conflicts with PortAudio dependencies.
- The sample app might crash if it's using libCurl 7.69.
Version 1.19.1
Release date
April 27, 2020
Bug fixes
- Fixed a bug that caused Display Cards for certain ExternalMediaPlayer (EMP) adapters to stop rendering.
Version 1.19.0
Release date
April 13 2020
Enhancements
- Added support for AudioPlayer Interface 1.4.
- Updated the ExternalMediaPlayer Interface. ExternalMediaPlayer adapters can now be integrated into the SDK without an overlay. Provide the paths of the adapter directories with
EXTERNAL_MEDIA_PLAYER_ADAPTER_PATHS
. Any adapters might require updating. For more details, refer to ExternalMediaPlayerAdapters.cmake. - Added three new diagnostic tools, allowing for file audio injection, protocol trace, and device properties aggregation. Audio injection replaces a real microphone and can't be used if another microphone, such as PortAudio, is enabled. Note that these tools are available in
DEBUG
mode only, and must not be enabled in devices built for release. - Added support for local ducking. This feature allows on-device content to attenuate the volume of other channels when necessary.
- Updated the following
SpeakerManagerInterface
APIs:setVolume
,adjustVolume
andsetMute
. These changes help you determine which components get notified when settings change (either AVS or SDK observers). - Added attenuation/ducking support to music.
- Updated
MediaPlayer::setSource
to specify the underlying file type that's passed with the raw data – for example, .mp3 or .wav. - Added new CMake parameters to build with customized versions of curl, SQlite, and Crypto.
Bug fixes
- Fixed issue that caused CBLAuthDelegateConfiguration.cpp file to read incorrect settings
- Fixed issue that caused a song to skip when rapidly pausing and playing music.
Known issues
- Build errors can occur on the Raspberry Pi due to incorrect linking of the atomic library. A suggested workaround is to add the following
set(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -latomic")
to the top most CMake file. - The WebVTT dependency required for captions isn't supported for Windows/Android.
- Exiting from the setting option takes you back to the Options Menu directly. It doesn't provide a message to indicate that you're back in the main menu.
- Failing Unit Tests and AIP Unit tests are disabled on Windows
- AudioInputProcessor unit tests don't build on Windows when with the
-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter. - Music playback history isn't displayed in the Alexa app for certain account and device types.
- The Gnu Compiler Collection 8+ (GCC 8+),
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this problem by upgrading to ADB version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned though local TTS.
- ACL encounters issues if it receives audio attachments but doesn't consume them.
- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- Interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This only impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
ServerDisconnectIntegratonTest
tests are disabled until they're updated to reflect new service behavior.- The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently. - Raspberry Pi sample app might crash after prolonged use due to conflicts with PortAudio dependencies.
- The sample app might crash if it's using libCurl 7.69.
Version 1.18.0
Release date
February 19, 2020
Enhancements
- Added support for Bluetooth Interface 2.0. This interface adds support for multiple simultaneous connections to Bluetooth peripherals.
- Added support for Audio Focus Manager Library (AFML) Multi Activity. This interface enhances the behavior of a device so it can handle more than one Activity per Channel.
- Added the
obfuscatePrivateData
logging method to help remove certain data from logs. - Updated
MediaPlayerObserverInterface
to include metadata about playback states. - Added SDK extension point. You can integrate CMake projects into the SDK without cloning those projects into a subdirectory.
Bug fixes
- Fixed Mac/OSX issue that caused an unresponsive sample app when not connected to the internet.
- Fixed issue that prevented sample app from exiting various states.
- Fixed
UIManager
issue that caused an error in the logs when the device with built without the wake word enabled. - Fixed volume issue that caused timers to ascend in volume when setting up ascending alarms.
- Fixed alert volume issue that caused any changes to the alert volume to notify observers.
- Fixed EQ issue where changes to the EQ band levels didn't notify observers.
Known issues
- Build errors can occur on the Raspberry Pi due to incorrect linking of the atomic library. A suggested workaround is to add the following
set(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -latomic")
to the top most CMake file. - The WebVTT dependency required for
captions
isn't supported for Windows/Android. - Exiting from the setting option takes you back to the Options Menu directly. It doesn't provide a message to indicate that you're back in the main menu.
- Failing Unit Tests and AIP Unit tests are disabled on Windows.
AudioInputProcessor
unit tests don't build on Windows when with the-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter.- Music playback history isn't displayed in the Alexa app for certain account and device types.
- The Gnu Compiler Collection 8+ (GCC 8+)
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this problem by upgrading to ADB version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned though local TTS.
- ACL encounters issues if it receives audio attachments but doesn't consume them.
- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- Interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This only impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
ServerDisconnectIntegratonTest
tests are disabled until they're updated to reflect new service behavior.- The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently.
2019
Version 1.17.0
Release date
December 10th 2019
Enhancements
- Added support for captions for TTS. This enhancement allows you to print on-screen captions for Alexa voice responses.
- Added support for SpeechSynthesizer Interface 1.3. This interface supports the new
captions
parameter. - Added support for AudioPlayer Interface 1.3. This interface supports the new
captions
parameter. - Added support for Interaction Model 1.2.
- Added support for System 2.0.
- Added support for Alerts Interface 1.4.
- Added support for Alarm Volume Ramp (Ascending Alarms in the Companion App). This feature lets the user enable alarm fade in. You enable this feature in the sample app through the settings menu.
- Added support to use certified senders for URI path extensions. This change allows you to specify the URI path extension when sending messages with
CertifiedSender::sendJSONMessage
- Added new
Metrics
interfaces and helper classes. These additions help you create and consumeMetrics
events.- Interfaces -
MetricRecorderInterface
,MetricSinkInterface
. - Helper Classes -
DataPointStringBuilder
,DataPointCounterBuilder
,DataPointDurationBuilder
,MetricEventBuilder
.
- Interfaces -
- Added support for the following AVS endpoint controller capabilities:
- Added
PowerResourceManagerInterface
. This interface allows the SDK to control power resource levels for components such as theAudioInputProcessor
andSpeechSynthesizer
. - Added
AlexaInterfaceCapabilityAgent
. This Capability Agent handles common directives and endpoint controller capabilities support byAlexa.AlexaInterface
. - Added
AlexaInterfaceMessageSenderInterface
. Use this interface to send common events defined by theAlexa.AlexaInterface
interface. - Added
BufferingComplete
toMediaPlayerObserverInterface
. This method helps improve performance in poor networking conditions by making sureMediaPlayer
pre-buffers correctly. - Added
SendDTMF
toCallManagerInterface
. This method allows you to send DTMF tones during calls.
New build options
- CAPTIONS
- ADDED
CAPTIONS
- ADDED
LIBWEBVTT_LIB_PATH
- ADDED
LIBWEBVTT_INCLUDE_DIR
- ADDED
- ENDPONTS
- ADDED
ENABLE_ALL_ENDPOINT_CONTROLLERS
- ADDED
ENABLE_POWER_CONTROLLER
- ADDED
ENABLE_TOGGLE_CONTROLLER
- ADDED
ENABLE_RANGE_CONTROLLER
- ADDED
ENABLE_MODE_CONTROLLER
- ADDED
New dependencies
- To use captions, you must install a new dependency – the libwebvtt parsing library. WebVTT is a C/C++ library for interpreting and authoring WebVTT content. WebVTT is a caption and subtitle format designed for use with HTML5 audio and video elements.
Bug fixes
- Fixed
MimeResponseSink::onReceiveNonMimeData
data issue that returned invalid data. - Fixed data type issue that incorrectly used
finalResponseCode
instead ofFinalResponseCodeId
- Fixed
UrlContentToAttachmentConverter
issue that used the incorrect range parameter. - Fixed
FinallyGuard
linking issue that caused problems compiling the SDK on iOS. - Fixed bug when you spoke the Wake Word "Alexa" twice rapidly.
Known issues
- The WebVTT dependency required for
captions
isn't supported for Windows/Android. AudioInputProcessor
unit tests don't build on Windows when with the-DCMAKE_BUILD_TYPE=DEBUG
CMake parameter.- Music playback history isn't displayed in the Alexa app for certain account and device types.
- When using Gnu Compiler Collection 8+ (GCC 8+),
-Wclass-memaccess
triggers warnings. You can ignore these, they don't cause the build to fail. - Android error
libDefaultClient.so not found
might occur. Resolve this by upgrading to ADB version 1.0.40. - If a device loses a network connection, the lost connection status isn't returned through local TTS.
- ACL encounters issues if it receives audio attachments but doesn't consume them.
SpeechSynthesizerState
usesGAINING_FOCUS
andLOSING_FOCUS
as a workaround for handling intermediate states.- Media streamed through Bluetooth might abruptly stop. To restart playback, resume the media in the source app or toggle next/previous.
- If a connected Bluetooth device is inactive, the Alexa app might indicates that audio is playing.
- The Bluetooth agent assumes that the Bluetooth adapter is always connected to a power source. Disconnecting from a power source during operation isn't yet supported.
- When using some products, interrupted Bluetooth playback might not resume if other content is locally streamed.
make integration
isn't available for Android. To run Android integration tests, manually upload the test binary and input file and run ADB.- Alexa might truncate the beginning of speech when responding to text-to-speech (TTS) user events. This only impacts Raspberry Pi devices running Android Things with HDMI output audio.
- A reminder TTS message doesn't play if the sample app restarts and loses a network connection. Instead, the default alarm tone plays twice.
ServerDisconnectIntegratonTest
tests are disabled until they're updated to reflect new service behavior.- The
DirectiveSequencerTest.test_handleBlockingThenImmediatelyThenNonBockingOnSameDialogId
test fails intermittently. - On some devices, Pressing
t
andh
in the sample app doesn't exit the assigned state. - Exiting the settings menu doesn't provide a message to indicate that you're back in the main menu.
Version 1.16.0
Release date
October 25th 2019
Enhancements
- Added support for SpeechSynthesizer version 1.2 which includes the new
playBehaviour
directive. - Added support for pre-buffering in the
AudioPlayer
Capability Agent. You can optionally choose the number of instancesMediaPlayer
uses in the AlexaClientSDKconfig.json. Important: the contract forMediaPlayerInterface
has changed. You must now make sure that theSourceId
value returned bysetSource()
is unique across all instances. - The
AudioPlayer
Capability Agent is now licensed under the Amazon Software License instead of the Apache Software License.
Bug fixes
- Fixed Android issue that caused the build script to ignore
PKG_CONFIG_PATH
. This sometimes caused the build to use a preinstalled dependency instead of the specific version downloaded by the Android script. For example openssl). - Fixed Android issue that prevented the sample app from running at the same time as other apps using the microphone. Android doesn't inherently allow two apps to use the microphone. Pressing the mute button now temporarily stops Alexa from accessing the microphone.
- Added 'quit' (– q) to the settings sub menu.
- Fixed outdated dependencies issue in the Windows install script.
- Fixed reminders issue that caused Notification LEDs to stay on, even after dismissing the alert.
For more details, check this.
Version 1.15.0
Release date
September 25th 2019
Enhancements
- Added
SystemSoundPlayer
toApplicationUtilities
.SystemSoundPlayer
is a new class that plays pre-defined sounds. Supported sounds include the wake word notification and the end of speech tone. - Removed Echo Spatial Perception (ESP) functionality from the Alexa Voice Service (AVS) device SDK. Make sure you download and test your devices using the new AVS SDK sample app. If you're using an older version of the sample app, manually remove any references to ESP or errors occur during compile.
- Added
onNotificationReceived
toNotificationsObserverInterface
.onNotificationReceived
broadcasts whenNotificationsObserverInterface
receives a new notification, instead of only sending the indicator state. This is important if you support a feature that requires a distinct signal for each notification received. SeeNotificationsObserverInterface
for more details. - Added support for Multilingual Mode . With this enabled, Alexa automatically detects what language a user speaks by analyzing the spoken wake word and proceeding utterances. After Alexa identifies the language, all corresponding responses are in the same language. The current supported language pairs are:
[ "en-US", "es-US" ]
[ "es-US", "en-US" ]
[ "en-IN", "hi-IN" ]
[ "hi-IN", "en-IN" ]
[ "en-CA", "fr-CA" ]
[ "fr-CA", "en-CA" ]
IMPORTANT: Specify the locales your device supports in thelocaleCombinations
field inAlexaClientSDKConfig.json
. This field can't be empty. If you don't set these values, the sample app fails to run.- Added two new system settings,
Timezone
andLocale
.- Timezone: For example, you can set the
defaultTimezone
toAmerica/Vancouver
. If you don't set a value,GMT
is set as the default value. If you set a new timezone, make sure that your AVS system settings and default time zone stay in sync. To handle this, use the new classSystemTimeZoneInterface
. - Locale: For example, you can set
defaultLocale
toen-GB
, instead of the defaulten-US
.
- Timezone: For example, you can set the
- The
SpeechRecognizer
interface now supports the following functionalities.- Change wake word (
Alexa
supported for now). - Toggle start of request tone on/off.
- Toggle End of request tone on/off.
- Change wake word (
- Deprecated the
CapabilityAgents
Settings{…}
library.Settings {…}
now maps to an interface that's no longer supported. You might need to update your code to handle these changes. Read Settings Interface for more details. - Added support for three new locals: Spanish - United States (ES_US), Hindi - India (HI_IN), and Brazilian - Portuguese (PT_BR).
- Linked the atomic library to the sample app to prevent build errors on Raspberry Pi.
Bug fixes
- Fixed resource leaking in EqualizerCapabilityAgent after engine shutdown.
- Issue 1391: Fixed an issue where SQLiteDeviceSettingsStorage::open tries to acquire a mutex twice, resulting in deadlock.
- Issue 1468: Fixed a bug in AudioPlayer::cancelDirective that causes a crash.
- Fixed Windows install script that caused the sample app build to fail. removed pip, flask, requests, and commentjson dependencies from the mingw.sh helper script.
- Fixed issue: notifications failed to sync upon device initialization. For example, let's say you had two devices. one turned on and the other turned off. After clearing the notification on the first device, it still showed up on the second device after turning it on.
- Fixed issue: barging in on a reminder caused it to stick in an inconsistent state, blocking subsequent reminders. For example, if a reminder was going off and you interrupted it, the reminder would get persist indefinitely. You could schedule future reminders, but they wouldn't play. Saying “Alexa stop” or rebooting the device fixed the “stuck” reminder.
Version 1.14.0
Release date
July 09 2019
Enhancements
- AudioPlayer can now pre-buffer audio tracks in the Pre-Handle stage.
Bug fixes
- Fixed an issue in the SQLite wrapper code where a
SQLiteStatement
caused a memory corruption issue. - Fixed a race condition in SpeechSynthesizer that caused crashes.
- Fixed a
cmake
issue that specifies a dependency for Bluetooth incorrectly. - Fixed a bug that caused Bluetooth playback to start automatically.
- Changed
supportedOperations
from a vector to a set inExternalMediaAdapterInterface
. - Corrected an issue where a
VolumeChanged
event had previously been sent when the volume was unchanged aftersetVolume
oradjustVolume
had been called locally. - Fixed issue with
IterativePlaylistParser
that prevented live stations on TuneIn from playing on Android. - Corrected the spelling of "UNINITIALIZED".
Version 1.13.0
Release date
May 22 2019
Enhancements
- When an active Alert moves to the background, the alert now begins after a 10-second delay. Alert loop iteration delays can now no longer last longer than a maximum of 10 seconds, rather than depending on the length of the audio asset.
- Changed NotificationsSpeaker to use Alerts Volume instead of using the speaker volume.
- Allow customers to pass in an implementation of InternetConnectionMonitorInterface which will force AVSConnectionManager to reconnect on internet connectivity loss.
- Added an exponential wait time for retrying transmitting a message via CertifiedSender.
- When Volume is set to 0 and device is unmuted, volume is bumped up to a non-zero value. When Volume is set to 0 and Alexa talks back to you, volume is bumped up to a non-zero value.
- Deprecated HttpResponseCodes.h, which is now present only to ensure backward compatibility.
- The default base URLs for AVS have changed. These new URLs are supported by SDK v1.13 and later versions. Amazon recommends that all new and existing implementations update to v1.13 or later and use the new base URLs accordingly; however, Amazon will continue to support the legacy base URLs.
Bug fixes
- Fixed bug where receiving a Connected = true Property change from BlueZ without UUID information resulted in BlueZBluetoothDevice transitioning to CONNECTED state.
- Fixed bug where MediaStreamingStateChangedEvent may be sent on non-state related property changes.
- Added null check to SQLiteStatement::getColumnText.
- Fixed an issue where database values with unescaped single quotes passed to miscStorage database will fail to be stored. Added a note on the interface that only non-escaped values should be passed.
- Fixed a loop in audio in live stations based on playlists.
- Fixed a race condition in TemplateRuntime that may result in a crash.
- Fixed a race condition where a recognize event due to a EXPECT_SPEECH may end prematurely.
- Changed the name of Alerts channel to Alert channel within AudioActivityTracker.
- Prevented STOP Wakeword detections from generating Recognize events.
- The SQLiteDeviceSettingsStorageTest no longer fails for Android.
Version 1.12.1
Release date
April 02 2019
Bug fixes
- Fixed a bug where the same URL was being requested twice when streaming iHeartRadio. Now, a single request is sent.
- Corrected pause/resume handling in
ProgressTimer
so that extraProgressReportDelayElapsed
events are not sent to AVS.
Version 1.12.0
Release date
February 25 2019
Enhancements
- Support was added for the
fr_CA
locale. - The Executor has been optimized to run a single thread when there are active job in the queue, and to remain idle when there are not active jobs.
- An additional parameter of
alertType
has been added to the Alerts capability agent. This will allow observers of alerts to know the type of alert being delivered. - Support for programmatic unload and load of PulseAudio Bluetooth modules was added. To enable this feature, there is a new CMake parameter :
BLUETOOTH_BLUEZ_PULSEAUDIO_OVERRIDE_ENDPOINTS
. Note that libpulse-dev is a required dependency of this feature. - An observer interface was added for when an active Bluetooth device connects and disconnects.
- The
BluetoothDeviceManagerInterface
instantiation was moved fromDefaultClient
toSampleApp
to allow apps to override it. - The
MediaPlayerInterface
now supports repeating playback of URL sources. - The Kitt.AI wake word engine (WWE) is now compatible with GCC5+.
- Stop of ongoing alerts, management of MessageObservers, and management of CallStateObservers have been exposed through DefaultClient.
Bug fixes
- Issue 953 - The
MediaPlayerInterface
requirement that callbacks not be made upon a callers thread has been removed. - Issue 1136 - Added a missing default virtual destructor.
- Issue 1140 - Fixed an issue where DND states were not synchronized to the AVS cloud after device reset.
- Issue 1143 - Fixed an issue in which the SpeechSynthesizer couldn't enter a sleeping state.
- Issue 1183 - Fixed an issue where alarm is not sounding for certain time zones
- Changing an alert's volume from the Alexa app now works when an alert is playing.
- Added missing shutdown handling for ContentDecrypter to prevent the
Stop
command from triggering a crash when SAMPLE-AES encrypted content was streaming. - Fixed a bug where if the Notifications database is empty, due to a crash or corruption, the SDK initialization process enters an infinite loop when it retries to get context from the Notifications capability agent.
- Fixed a race condition that caused
AlertsRenderer
observers to miss notification that an alert has been completed.
2018
Version 1.11.0
Release date
December 19 2018
Enhancements
- Added support for the new Alexa DoNotDisturb interface, which enables users to toggle the do not disturb (DND) function on their Alexa built-in products.
- The SDK now supports Opus encoding, which is optional. To enable Opus, you must set the CMake flag to
-DOPUS=ON
, and include the libopus library dependency in your build. - The MediaPlayer reference implementation has been expanded to support the SAMPLE-AES and AES-128 encryption methods for HLS streaming.
- AES-128 encryption is dependent on libcrypto, which is part of the required openSSL library, and is enabled by default.
- To enable SAMPLE-AES encryption, you must set the
-DSAMPLE_AES=ON
in your CMake command, and include the FFMPEG library dependency in your build.
- A new configuration for deviceSettings has been added. This configuration allows you to specify the location of the device settings database.
- Added locale support for es-MX.
Bug fixes
- Fixed an issue where music wouldn't resume playback in the Android app.
- Now all equalizer capabilities are fully disabled when equalizer is turned off in configuration file. Previously, devices were unconditionally required to provide support for equalizer in order to run the SDK.
- Issue 1106 - Fixed an issue in which the
CBLAuthDelegate
wasn't using the correct timeout during request refresh. - Issue 1128 - Fixed an issue in which the
AudioPlayer
instance persisted at shutdown, due to a shared dependency with theProgressTimer
. - Fixed in issue that occurred when a connection to streaming content was interrupted, the SDK did not attempt to resume the connection, and appeared to assume that the content had been fully downloaded. This triggered the next track to be played, assuming it was a playlist.
- Issue 1040 - Fixed an issue where alarms would continue to play after user barge-in.
Version 1.10.0
Release date
October 24 2018
Enhancements
- New optional configuration for EqualizerController. The EqualizerController interface allows you to adjust equalizer settings on your product, such as decibel (dB) levels and modes.
- Added reference implementation of the EqualizerController for GStreamer-based (MacOS, Linux, and Raspberry Pi) and OpenSL ES-based (Android) MediaPlayers. Note: In order to use with Android, it must support OpenSSL ES.
- New optional configuration for the TemplateRuntime display card value.
- A configuration file generator script,
genConfig.sh
is now included with the SDK in the tools/Install directory.genConfig.sh
and it's associated arguments populateAlexaClientSDKConfig.json
with the data required to authorize with LWA. - Added Bluetooth A2DP source and AVRCP target support for Linux.
- Added Alexa for Business (A4B) support, which includes support for handling the new RevokeAuthorization directive in the Settings interface. A new CMake parameter has been added to enable A4B within the SDK,
-DA4B
. - Added locale support for IT and ES.
- The Alexa Communication Library (ACL),
CBLAUthDelegate
, and sample app have been enhanced to detect de-authorization using the newz
command. - Added
ExternalMediaPlayerObserver
, which receives notification of player state, track, and username changes. HTTP2ConnectionInterface
was factored out ofHTTP2Transport
to enable unit testing ofHTTP2Transport
and re-use ofHTTP2Connection
logic.
Bug fixes
- Fixed a bug in which
ExternalMediaPlayer
adapter playback wasn't being recognized by AVS. - Issue 973 - Fixed issues related to
AudioPlayer
where progress reports were being sent out of order or with incorrect offsets. - An
EXPECTING
, state has been added toDialogUXState
in order to handleEXPECT_SPEECH
state for hold-to-talk devices. - Issue 948 - Fixed a bug in which the sample app got stuck in various states.
- Fixed a bug where there was a delay between receiving a
DeleteAlert
directive, and deleting the alert. - Issue 839 - Fixed an issue where speech was being truncated due to the
DialogUXStateAggregator
transitioning between aTHINKING
andIDLE
state. - Fixed a bug in which the
AudioPlayer
attempted to play when it wasn't in theFOREGROUND
focus. CapabilitiesDelegateTest
now works on Android.- Issue 950 - Improved Android Media Player audio quality.
- Issue 908 - Fixed compile error on g++ 7.x in which includes were missing.
Version 1.9.0
Release date
August 28 2018
Enhancements
- Added Android SDK support, which includes new implementations of the MediaPlayer, audio recorder, and logger.
- Added the InteractionModel interface, which enables Alexa Routines.
- Optional configuration changes have been introduced. Now a network interface can be specified to connect to the SDK via curl.
- Build options can be configured to support Android.
- Added GUI 1.1 support. The
PlaybackController
has been extended to support new control functionality, and theSystem
interface has been updated to supportSoftwareInfo
.
Bug fixes
- Installation script execution time has been reduced. Now a single branch clone is used, such as the main branch.
- Issue 846 - Fixed a bug where audio stuttered on slow network connections.
- Removed the
SpeakerManager
constructor check for non-zero speakers. - Issue 891 - Resolved incorrect offset in the
PlaybackFinished
event. - Issue 727 - Fixed an issue where the sample app behaved erratically upon network disconnection/reconnection.
- Issue 910 - Fixed a GCC 8+ compilation issue. Note: issues related to
-Wclass-memaccess
will still trigger warnings, but won't fail compilation. - Issue 871 Issue 880 - Fixed compiler warnings.
- Fixed an issue where
PlaybackStutterStarted
andPlaybackStutterFinished
events were not being sent due to a missing Gstreamer queue element. - Fixed a bug where the
CapabilitiesDelegate
database was not being cleared upon logout. - Fixed in issue that caused the following compiler warning “class has virtual functions but non-virtual destructor”.
- Fixed a bug where
BlueZDeviceManager
was not properly destroyed. - Fixed a bug that occurred when the initializer list was converted to
std::unordered_set
. - Fixed a build error that occurred when building with
BUILD_TESTING=Off
.
Version 1.8.1
Release date
July 09 2018
Enhancements
- Added support for adjustment of alert volume.
- Added support for deletion of multiple alerts.
- The following
SpeakerInterface::Type
enumeration values have changed:AVS_SYNCED
is nowAVS_SPEAKER_VOLUME
.LOCAL
is nowAVS_ALERTS_VOLUME
.
Version 1.8.0
Release date
June 27 2018
Enhancements
- Added local stop functionality. This allows a user to stop an active function, such as an alert or timer, by uttering "Alexa, stop" when an Alexa-enabled product is offline.
- Alerts in the background now stream in 10 sec intervals, rather than continuously.
- Added support for France to the sample app.
- Updated the ACL MIME type for sending JSON to AVS from
"text/json"
to"application/json"
. friendlyName
can now be updated forBlueZ
implementations ofBlueZBluetoothDevice
andBlueZHostController
.
Bug fixes
- Fixed an issue where the Bluetooth agent didn't clear user data upon reset, including paired devices and the
uuidMapping
table. - Fixed
MediaPlayer
threading issues. Now each instance has it's ownglib
main loop thread, rather than utilizing the default main context worker thread. - Fixed segmentation fault issues that occurred when certain static initializers needed to be initialized in a certain order, but the order wasn't defined.
Version 1.7.1
Release date
May 04 2018
Enhancements
- Added the Bluetooth interface, which manages the Bluetooth connection between Alexa-enabled products and peer devices. This release supports
A2DP-SINK
andAVRCP
profiles. Note: Bluetooth is optional and is currently limited to Raspberry Pi and Linux platforms. - Added new Bluetooth dependencies for Linux and Raspberry Pi.
- Device Capability Framework (
DCF
) renamed toCapabilities
. - Updated the non-CBL client ID error message to be more specific.
- Updated the sample app to enter a limited interaction mode after an unrecoverable error.
Bug fixes
- Issue 597 - Fixed a bug where the sample app did not respond to locale change settings.
- Fixed issue where GStreamer 1.14
MediaPlayerTest
failed on Windows. - Fixed an issue where a segmentation fault was triggered after unrecoverable error handling.
Version 1.7.0
Release date
April 18 2018
Enhancements
AuthDelegate
andAuthServer.py
have been replaced byCBLAUthDelegate
, which uses Code Based Linking for authorization.- Added new properties to
AlexaClientSDKConfig
:cblAuthDelegate
- This object specifies parameters forCBLAuthDelegate
.miscDatabase
- A generic key/value database to be used by various components.dcfDelegate
- This object specifies parameters forDCFDelegate
. Within this object, values were added forendpoint
andoverridenDcfPublishMessageBody
.endpoint
is the endpoint for the Capabilities API.overridenDcfPublishMessageBody
is the message that is sent to the Capabilities API. Note: Values in thedcfDelegate
object will only work inDEBUG
builds.deviceInfo
- Specifies device-identifying information for use by the Capabilities API andCBLAuthDelegate
.
- Updated Directive Sequencer to support wildcard directive handlers. This allows a handler for a given AVS interface to register at the namespace level, rather than specifying the names of all directives within a given namespace.
- Updated the Raspberry Pi installation script to include
alsasink
inAlexaClientSDKConfig
. - Added
audioSink
as a configuration option. This allows users to override the audio sink element used inGstreamer
. - Added an interface for monitoring internet connection status:
InternetConnectionMonitorInterface.h
. - The Alexa Communications Library (ACL) is no longer required to wait until authorization has succeeded before attempting to connect to AVS. Instead,
HTTP2Transport
handles waiting for authorization to complete. - Device capabilities can now be sent for each capability interface using the Capabilities API.
- The sample app has been updated to send Capabilities API messages, which are automatically sent when the sample app starts. Note: A successful call to the Capabilities API must occur before a connection with AVS is established.
- The SDK now supports HTTP PUT messages.
- Added support for opt-arg style arguments and multiple configuration files. Now, the sample app can be invoked by either of these commands:
SampleApp <configfile> <debuglevel>
ORSampleApp -C file1 -C file2 ... -L loglevel
.
Bug fixes
- Fixed Issues 447 and 553 .
- Fixed the
AttachmentRenderSource
's handling ofBLOCKING
AttachmentReaders
. - Updated the
Logger
implementation to be more resilient tonullptr
string inputs. - Fixed a
TimeUtils
utility-related compile issue. - Fixed a bug in which alerts failed to activate if the system was restarted without network connection.
- Fixed Android 64-bit build failure issue.
Version 1.6.0
Release date
March 08, 2018
Enhancements
rapidJson
is now included with "make install".- Updated the
TemplateRuntimeObserverInterface
to support clearing ofdisplayCards
. - Added Windows SDK support, along with an installation script (MinGW-w64).
- Updated
ContextManager
to ignore context reported by a state provider. - The
SharedDataStream
object is now associated by playlist, rather than by URL. - Added the
RegistrationManager
component. Now, when a user logs out all persistent user-specific data is cleared from the SDK. The log out functionality can be exercised in the sample app with the new command:k
.
Bug fixes
- Issue 400 Fixed a bug where the alert reminder did not iterate as intended after loss of network connection.
- Issue 477 Fixed a bug in which Alexa's weather response was being truncated.
- Fixed an issue in which there were reports of instability related to the Sensory engine. To correct this, the
portAudio
suggestedLatency
value can now be configured.
Version 1.5.0
Release date
February 12 2018
Enhancements
- Added the
ExternalMediaPlayer
Capability Agent. This allows playback from music providers that control their own playback queue. Example: Spotify. - Added support for AU and NZ to the
SampleApp
. - Firmware version can now be sent to Alexa via the
SoftwareInfo
event. The firmware version is specified in the config file under thesampleApp
object as an integer value namedfirmwareVersion
. - The new
f
command was added to theSampleApp
which allows the firmware version to be updated at run-time. - Optional configuration changes have been introduced. Now a default log level can be set for
ACSDK_LOG_MODULE
components, globally or individually. This value is specified under a new root level configuration object calledlogger
, and the value itself is namedlogLevel
. This allows you to limit the degree of logging to that default value, such asERROR
orINFO
.
Bug fixes
- Fixed bug where
AudioPlayer
progress reports were not being sent, or were being sent incorrectly. - Issue 408 - Irrelevant code related to
UrlSource
was removed from theGStreamer-based MediaPlayer
implementation. - The
TZ
variable no longer needs to be set toUTC
when building theSampleApp
. - Fixed a bug where
CurlEasyHandleWrapper
logged unwanted data on failure conditions. - Fixed a bug to improve
SIGPIPE
handling. - Fixed a bug where the file name and class name were mismatched. Changed
UrlToAttachmentConverter.h
toUrlContentToAttachmentConverter.h
,andUrlToAttachmentConverter.cpp
toUrlContentToAttachmentConverter.cpp
. - Fixed a bug where after muting and then un-muting the GStreamer-based
MediaPlayer
implementation, the next item in queue would play instead of continuing playback of the originally muted item.
Version 1.4.0
Release date
January 12 2018
Enhancements
- Added the Notifications Capability Agent. This allows a client to receive notification indicators from Alexa.
- Added support for the
SoftwareInfo
event. This code is triggered in theSampleApp
by providing a positive decimal integer as the "firmwareVersion" value in "sampleApp" object of theAlexaClientSDKConfig.json
. The reported firmware version can be updated after starting theSampleApp
by callingSoftwareInfoSender::setFirmwareVersion()
. This code path can be exercised in theSampleApp
with the new command:f
. - Added unit tests for Alerts.
- The GStreamer-based pipeline allows for the configuration of
MediaPlayer
output based on information provided inConfig
. - Playlist streaming now uses a
BLOCKING
writer, which improves streaming efficiency.
Bug fixes
- Fixed bug where
SpeechSynthesizer
would not stop playback when a state change timeout was encountered. - Fixed the
SampleApplication
destructor to avoid segfaults if the object is not constructed correctly. - Fixed bug where
AudioPlayer
would erroneously callexecuteStop()
incancelDirective()
. - Issue 396 - Fixed bug for compilation error with GCC7 in
AVSCommon/SDKInterfaces/include/AVSCommon/SDKInterfaces/Audio/AlertsAudioFactoryInterface.h
- Issue 384 - Fixed bug that caused
AuthServer.py
to crash. - Fixed bug where a long delay was encountered after pausing and resuming a large Audible chapter.
- Fixed bug that caused named timers and reminders to loop for an additional
loopCount
. - Fixed memory corruption bug in
MessageInterpreter
. - Fixed illegal memory accesses in
MediaPlayer
logging.
2017
Version 1.3.0
Release date
December 08 2017
- Enhancements
- ContextManager now passes the namespace/name of the desired state to StateProviderInterface::provideState(). This is helpful when a single StateProvider object provides multiple states, and needs to know which one ContextManager is asking for.
- The mime parser was hardened against duplicate boundaries.
- Added functionality to add and remove AudioPlayer observers to the DefaultClient.
- Unit tests for Alerts were added.
- Added en-IN, en-CA and ja-JP to the SampleApp locale selection menu.
- Added default alert and timer audio data to the SDK SampleApp. There is no longer a requirement to download these audio files and configure them in the json configuration file.
- Added support in SDS Reader and AttachmentReader for seeking into the future. This allows a reader to move to an index which has not yet arrived in the SDS and poll/block until it arrives.
- Added support for blocking Writers in the SharedDataStream class.
- Changed the default status code sent by MessageRequestObserverInterface::onSendCompleted() to SERVER_OTHER_ERROR, and mapped HTTP code 500 to SERVER_INTERNAL_ERROR_V2.
- Added support for parsing stream duration out of playlists.
- Added a configuration option ("sampleApp":"displayCardsSupported") that allows the displaying of display cards to be enabled or disabled.
- Named Timers and Reminders have been updated to fall back to the on-device background audio sound when cloud urls cannot be accessed or rendered.
- Bug fixes
- Removed floating point dependencies from core SDK libraries.
- Fixed bug in SpeechSynthesizer where it's erroneously calling stop more than once.
- Fixed an issue in ContentFetcher where it could stop responding during destruction until an active GET request completed.
- Fixed a couple of parsing bugs in LibCurlHttpContentFetcher related to case-sensitivity and mime-type handling.
- Fixed a bug where MediaPlayerObserverInterface::onPlaybackResumed() wasn't being called after resuming from a pause with a pending play/resume.
- Fixed a bug in LibCurlContentFetcher where it could error out if data is written to the SDS faster than it is consumed.
- The GStreamer-based MediaPlayer reference implementation now uses the ACL HTTP configured client.
- An API change has been made to MediaPlayerInterface::setSource(). This method now takes in an optional offset as well to allow for immediately streaming to the offset if possible.
- Next and Previous buttons now work with Audible.
- Pandora resume stuttering is addressed.
- Pausing and resuming Amazon music no longer seeks back to the beginning of the song.
- libcurl CURLOPT_NOSIGNAL option is set to 1 (https://curl.haxx.se/libcurl/c/CURLOPT_NOSIGNAL.html) to avoid crashes observed in SampleApp.
- Fixed the timing of the PlaybackReportIntervalEvent and PlaybackReportDelayEvent as specified in the directives.
- Fixed potential deadlocks in MediaPlayer during shutdown related to queued callbacks.
- Fixed a crash in MediaPlayer that could occur if the network is disconnected during playback.
- Fixed a bug where music could keep playing while Alexa is speaking.
- Fixed a bug which was causing problems with pause/resume and next/previous with Amazon Music.
- Fixed a bug where music could briefly start playing between speaks.
- Fixed a bug where HLS playlists would stop streaming after the initial playlist had been played to completion.
- Fixed a bug where Audible playback could not advance to the next chapter.
- Fixed some occurrences of SDK entering the IDLE state during the transition between Listening and Speaking states.
- Fixed a bug where PlaybackFinished events were not reporting the correct offset.
- An API change has been made to MediaPlayerInterface::getOffset(). This method is now required to return the final offset when called after playback has stopped.
- Fixed a problem where AIP was erroneously resetting its state upon getting a cancelDirective() callback.
Version 1.2.1
Release date
November 11 2017
Enhancements
- Added comments to
AlexaClientSDKConfig.json
. These descriptions provide additional guidance for what is expected for each field. - Enabled pause and resume controls for Pandora.
Bug fixes
- Bug fix for issue #329 -
HTTP2Transport
instances no longer leak whenSERVER_SIDE_DISCONNECT
is encountered. - Bug fix for issue #189 - Fixed a race condition in the
Timer
class that sometimes causedSpeechSynthesizer
to get stuck in the "Speaking" state. - Bug fix for a race condition that caused
SpeechSynthesizer
to ignore subsequentSpeak
directives. - Bug fix for corrupted mime attachments.
Version 1.2.0
Release date
October 27 2017
Enhancements
- Updated MediaPlayer to solve stability issues
- All capability agents were refined to work with the updated MediaPlayer
- Added the TemplateRuntime capability agent
- Added the SpeakerManager capability agent
- Added a configuration option ("sampleApp":"endpoint") that allows the endpoint that SampleApp connects to be specified without changing code or rebuilding
- Added very verbose capture of libcurl debug information
- Added an observer interface to observer audio state changes from AudioPlayer
- Added support for StreamMetadataExtracted Event. Stream tags found in the stream are represented in JSON and sent to AVS
- Added to the SampleApp a simple GuiRenderer as an observer to the TemplateRuntime Capability Agent
- Moved shared libcurl functionality to AVSCommon/Utils
- Added a CMake parameter to exclude tests from the "make all" build. Use "cmake
-DACSDK_EXCLUDE_TEST_FROM_ALL=ON" to enable it. When this option is enabled "make unit" and "make integration" still could be used to build and run the tests
Bug fixes
- Previously scheduled alerts now play following a restart
- General stability fixes
- Bug fix for CertifiedSender blocking forever if the network goes down while it's trying to send a message to AVS
- Fixes for known issue of Alerts integration tests fail: AlertsTest.UserLongUnrelatedBargeInOnActiveTimer and AlertsTest.handleOneTimerWithVocalStop
- Attempting to end a tap-to-talk interaction with the tap-to-talk button wouldn't work
- SharedDataStream could encounter a race condition due to a combination of a blocking Reader and a Writer closing before writing any data
- Bug-fix for the ordering of notifications within alerts scheduling. This fixes the issue where a local-stop on an alert would also stop a subsequent alert if it were to begin without delay
Version 1.1.0
Release date
October 10 2017
Enhancements
- Better GStreamer error reporting. MediaPlayer used to only report
MEDIA_ERROR_UNKNOWN
, now reports more specific errors as defined inErrorType.h
. - Codebase has been formatted for easier reading.
DirectiveRouter::removeDirectiveHandler()
signature changed and now returns a boolean indicating if given handler should be successfully removed or not.- Cleanup of raw and shared pointers in the creation of
Transport
objects. HTTP2Stream
s now have IDs assigned as they're acquired as opposed to created, making associated logs easier to interpret.AlertsCapabilityAgent
has been refactored.- Alert management has been factored out into an
AlertScheduler
class.
- Alert management has been factored out into an
- Creation of Reminder (implements Alert) class.
- Added new capability agent for
PlaybackController
with unit tests. - Added Settings interface with unit tests.
- Return type of
getOffsetInMilliseconds()
changed fromint64_t
tostd::chronology::milliseconds
. - Added
AudioPlayer
unit tests. - Added teardown for all Integration tests except Alerts.
- Implemented PlaylistParser.
Bug fixes
- AIP gets stuck in between states and refuses user input on network outage.
- SampleApp crashing if running for 5 minutes after network disconnect.
- Issue where on repeated user barge-ins,
AudioPlayer
would not pause. Specifically, the third attempt to “Play iHeartRadio” would not result in currently-playing music pausing. - Utterances being ignored after particularly long TTS.
- GStreamer errors cropping up on SampleApp exit as a result of accessing the pipeline before it’s been setup.
- Crashing when playing one URL after another.
- Buffer overrun in Alerts Renderer.
- SampleApp crashing when issuing "Alexa skip" command with iHeartRadio.
HTTP2Transport
network thread triggering a join on itself.HTTP2Stream
request handling truncating exception messages.AudioPlayer
was attempting an incorrect state transition fromSTOPPED
toPLAYING
through aplaybackResumed
.
Version 1.0.3
Release date
October 10 2017
- Enhancements
- Implemented
setOffSet
inMediaPlayer
. - Updated
LoggerUtils.cpp
.
- Implemented
- Bug fixes
- Bug fix to address incorrect stop behavior caused when Audio Focus is set to
NONE
and released . - Bug fix for intermittent failure in
handleMultipleConsecutiveSpeaks
. - Bug fix for
jsonArrayExist
incorrectly parsing JSON when trying to locate array children. - Bug fix for ADSL test failures with
sendDirectiveWithoutADialogRequestId
. - Bug fix for
SpeechSynthesizer
showing the wrong UX state when a burst ofSpeak
directives are received. - Bug fix for recursive loop in
AudioPlayer.Stop
.
- Bug fix to address incorrect stop behavior caused when Audio Focus is set to
Version 1.0.2
Release date
August 23 2017
Updates
- Removed code from AIP which propagates ExpectSpeech initiator strings to subsequent Recognize events. This code will be re-introduced when AVS starts sending initiator strings.
Version 1.0.1
Release date
August 08 2017
Updates
- Added a fix to the sample app so that the
StateSynchronization
event is the first that gets sent to AVS. - Added a
POST_CONNECTED
enum toConnectionStatusObserver
. StateSychronizer
now automatically sends aStateSynchronization
event when it receives a notification thatACL
isCONNECTED
.- Added
make install
for installing the AVS Device SDK. - Added an optional
make networkIntegration
for integration tests for slow network (only on Linux platforms). - Added shutdown management which fully cleans up SDK objects during teardown.
- Fixed an issue with
AudioPlayer
barge-in which was preventing subsequent audio from playing. - Changed
Mediaplayer
buffering to reduce stuttering. - Known Issues:
- Connection loss during some states keep the app in that state even after connection is regained. Pressing ‘s’ unsticks the state.
- Play/Pause media restarts it from the beginning.
SpeechSynthesizer
shows wrong UX state during a burst of Speaks.- Quitting the sample app while
AudioPlayer
is playing something causes a segmentation fault. AudioPlayer
sendingPlaybackPaused
during flash briefing.- Long Delay playing live stations on iHeartRadio.
- Teardown warnings at the end of integration tests.
Version 1.0.0
Release date
August 07 2017
Updates
- Added
AudioPlayer
capability agent.- Supports iHeartRadio.
StateSynchronizer
has been updated to better enforce thatSystem.SynchronizeState
is the first Event sent on a connection to AVS.- Additional tests have been added to
ACL
. - The
sample app
has been updated with several small fixes and improvements. ADSL
was updated such that all directives are now blocked while the handling of previousSpeechSynthesizer.Speak
directives complete. Because any directive may now be blocked, thepreHandleDirective(/ handleDirective()
path is now used for handling all directives.- Fixes for the following GitHub issues:
- A bug causing
ACL
to not send a ping to AVS every 5 minutes, leading to periodic server disconnects, was fixed. - Subtle race condition issues were addressed in the
Executor
class, resolving some intermittent crashes. - Known Issues
- Native components for the following capability agents are not included in this release:
PlaybackController
,Speaker
,Settings
,TemplateRuntime
, andNotifications
. MediaPlayer
- Long periods of buffer underrun can cause an error related with seeking and subsequent stopped playback.
- Long periods of buffer underrun can cause flip flopping between buffer_underrun and playing states.
- Playlist parsing is not supported unless -DTOTEM_PLPARSER=ON is specified.
AudioPlayer
- Amazon Music, TuneIn, and SiriusXM are not supported in this release.
- Our parsing of urls currently depends upon GNOME/totem-pl-parser which only works on some Linux platforms.
AlertsCapabilityAgent
- Satisfies the AVS specification except for sending retrospective Events. For example, sending
AlertStarted
Event for an Alert which rendered when there was no internet connection.
- Satisfies the AVS specification except for sending retrospective Events. For example, sending
sample app
:- Any connection loss during some states keeps the app stuck in that state, unless the ongoing interaction is manually stopped by the user.
- The user must wait several seconds after starting up the sample app before the sample app is properly usable.
Tests
:SpeechSynthesizer
unit tests stop responding on some older versions of GCC due to a tear down issue in the test suite- Intermittent Alerts integration test failures caused by rigidness in expected behavior in the tests
- Native components for the following capability agents are not included in this release:
Version 0.6.0
Release date
July 14 2017
Updates
- Added a sample app that leverages the SDK.
- Added
Alerts
capability agent. - Added the
DefaultClient
class. - Added the following classes to support directives and events in the
System
interface:StateSynchronizer
,EndpointHandler
, andExceptionEncounteredSender
. - Added unit tests for
ACL
. - Updated
MediaPlayer
to play local files given anstd::istream
. - Changed build configuration from
Debug
toRelease
. - Removed
DeprecatedLogger
class. - Known Issues:
MediaPlayer
: OurGStreamer
based implementation ofMediaPlayer
is not fully robust, and may result in fatal runtime errors, under the following conditions:- Attempting to play multiple simultaneous audio streams
- Calling
MediaPlayer::play()
andMediaPlayer::stop()
when the MediaPlayer is already playing or stopped, respectively. - Other miscellaneous issues, which will be addressed in the near future
AlertsCapabilityAgent
:- This component has been temporarily simplified to work around the known
MediaPlayer
issues mentioned above - Fully satisfies the AVS specification except for sending retrospective Events, for example, sending
AlertStarted
for an Alert which rendered when there was no Internet connection - This component is not fully thread-safe, however, this will be addressed shortly
- Alerts currently run indefinitely until stopped manually by the user. This will be addressed shortly by having a timeout value for an alert to stop playing.
- Alerts do not play in the background when Alexa is speaking, but will continue playing after Alexa stops speaking.
- This component has been temporarily simplified to work around the known
sample app
:- Without the refresh token being filled out in the JSON file, the sample app crashes on start up.
- Any connection loss during some states keeps the app stuck in that state, unless the ongoing interaction is manually stopped by the user.
- At the end of a shopping list with more than 5 items, the interaction in which Alexa asks the user if he/she would like to hear more does not finish properly.
Tests
:SpeechSynthesizer
unit tests stop responding on some older versions of GCC due to a tear down issue in the test suite- Intermittent Alerts integration test failures caused by rigidness in expected behavior in the tests
Version 0.5.0
Release date
June 23 2017
Updates
- Updated most SDK components to use new logging abstraction.
- Added a
getConfiguration()
method toDirectiveHandlerInterface
to register capability agents with Directive Sequencer. - Added
ACL
stream processing with pause and redrive. - Removed the dependency of
ACL
library onAuthdelegate
. - Added an interface to allow
ACL
to add/removeConnectionStatusObserverInterface
. - Fixed compile errors in KITT.ai,
DirectiveHandler
and compiler warnings inAIP
tests. - Corrected formatting of code in many files.
- Fixes for the following GitHub issues:
Version 0.4.1
Release date
June 9 2017
Updates
- Implemented Sensory wake word detector functionality.
- Removed the need for a
std::recursive_mutex
inMessageRouter
. - Added
AIP
unit tests. - Added
handleDirectiveImmediately
functionality toSpeechSynthesizer
. - Added memory profiles for:
- AIP
- SpeechSynthesizer
- ContextManager
- AVSUtils
- AVSCommon
- Bug fix for
MessageRouterTest
aborting intermittently. - Bug fix for
MultipartParser.h
compiler warning. - Suppression of sensitive log data even in debug builds. Use CMake parameter -DACSDK_EMIT_SENSITIVE_LOGS=ON to allow logging of sensitive information in DEBUG builds.
- Fixed crash in
ACL
when attempting to use more than 10 streams. - Updated
MediaPlayer
to useautoaudiosink
instead of requiringpulseaudio
. - Updated
MediaPlayer
build to suppport local builds of GStreamer. - Fixes for the following GitHub issues:
- MessageRouter::send() does not take the m_connectionMutex
- MessageRouter::disconnectAllTransportsLocked flow leads to erase while iterating transports vector
- Build errors when building with KittAi enabled
- HTTP2Transport race may lead to deadlock
- Crash in HTTP2Transport::cleanupFinishedStreams()
- The attachment writer interface should take a
const void*
instead ofvoid*
Version 0.4.0
Release date
May 31 2017 (patch)
Updates
- Added
AuthServer
, an authorization server implementation used to retrieve refresh tokens from LWA.
Version 0.4.0
Release date
May 24 2017
Updates
- Added
SpeechSynthesizer
, an implementation of theSpeechRecognizer
capability agent. - Implemented a reference
MediaPlayer
based on GStreamer for audio playback. - Added
MediaPlayerInterface
that allows you to implement your own media player. - Updated
ACL
to support asynchronous receipt of audio attachments from AVS. - Bug Fixes:
- Some intermittent unit test failures were fixed.
- Known Issues:
ACL
's asynchronous receipt of audio attachments may manage resources poorly in scenarios where attachments are received but not consumed.- When an
AttachmentReader
does not deliver data for prolonged periodsMediaPlayer
may not resume playing the delayed audio.
Version 0.3.0
Release date
May 17 2017
Updates
- Added the
CapabilityAgent
base class that is used to build capability agent implementations. - Added
ContextManager
, a component that allows multiple capability agents to store and access state. These Events includeContext
, which is used to communicate the state of each capability agent to AVS in the following Events: - Added
SharedDataStream
(SDS) to asynchronously communicate data between a local reader and writer. - Added
AudioInputProcessor
(AIP), an implementation of aSpeechRecognizer
capability agent. - Added WakeWord Detector (WWD), which recognizes keywords in audio streams. [0.3.0] implements a wrapper for KITT.ai.
- Added a new implementation of
AttachmentManager
and associated classes for use with SDS. - Updated
ACL
to support asynchronously sending audio to AVS.
Version 0.2.1
Release date
May 03 2017
Updates
- Replaced the configuration file
AuthDelegate.config
withAlexaClientSDKConfig.json
. - Added the ability to specify a
CURLOPT_CAPATH
value to be used when libcurl is used by ACL and AuthDelegate. See See Appendix C in the README for details. - Changes to ADSL interfaces:
- The [0.2.0] interface for registering directive handlers (
DirectiveSequencer::setDirectiveHandlers()
) was problematic because it canceled the ongoing processing of directives and dropped further directives until it completed. The revised API makes the operation immediate without canceling or dropping any handling. However, it does create the possibility thatDirectiveHandlerInterface
methodspreHandleDirective()
andhandleDirective()
may be called on different handlers for the same directive. DirectiveSequencerInterface::setDirectiveHandlers()
was replaced byaddDirectiveHandlers()
andremoveDirectiveHandlers()
.DirectiveHandlerInterface::shutdown()
was replaced withonDeregistered()
.DirectiveHandlerInterface::preHandleDirective()
now takes astd::unique_ptr
instead of astd::shared_ptr
toDirectiveHandlerResultInterface
.DirectiveHandlerInterface::handleDirective()
now returns a bool indicating if the handler recognizes themessageId
.
- The [0.2.0] interface for registering directive handlers (
- Bug fixes:
- ACL and AuthDelegate now require TLSv1.2.
onDirective()
now sendsExceptionEncountered
for unhandled directives.DirectiveSequencer::shutdown()
no longer sendsExceptionEncountered()
for queued directives.
Version 0.2.0
Release date
March 27 2017 (patch)
Updates
- Added memory profiling for ACL and ADSL. See Appendix A in the README.
- Added a command to build the API documentation.
Version 0.2.0
Release date
March 19 2017
Updates
- Added
Alexa Directive Sequencer Library
(ADSL) andAlexa Focus Manager Library
(AMFL). - CMake build types and options have been updated.
- Documentation for libcurl optimization included.
Version 0.1.0
Release date
-
February 10 2017
-
Initial release of the
Alexa Communications Library
(ACL), a component which manages network connectivity with AVS, andAuthDelegate
, a component which handles user authorization with AVS.