Sample App using AVS SDK about avs-device-sdk HOT 16 CLOSED

Hi Abhijit,

It sounds like you are on the right path here. You will need to develop some glue code which reads from your microphone interface and writes the audio data into a SharedDataStream. Your AudioProvider will consist of a pointer to your SharedDataStream, and additional metadata describing the stream. You can then make calls to AudioInputProcessor::recognize() with your AudioProvider, and the SDK will take care of generating the Recognize event and sending it out to AVS.

If you haven't already done so, I would suggest working through running the integration tests included with the SDK to work out any issues with your credentials.

Ken

from avs-device-sdk.

Thanks Ken. I will follow this.

Yes I did run Integration Tests. I am not facing any problems with credentials. Only one test I am seeing as failing. I have to look at it in more details to find out what exactly could be wrong.

from avs-device-sdk.

Hi @gabhijit-shorelineiot,

Do you have any other questions? Are you still seeing the one test failure?

Thanks,
Jamie

from avs-device-sdk.

Hi @JamieMeyers

Sorry I forgot to update. I keep seeing the following failure intermittently (though not always)

[ FAILED ] AlexaCommunicationsLibraryTest.testSendEventsSerially
[ FAILED ] AlexaCommunicationsLibraryTest.testSendEventsConcurrently
[ FAILED ] AlexaCommunicationsLibraryTest.testPersistentConnection

This doesn't look like Authenticaation issue, but some timing issue. There have been occasions when these tests are passed as well. (One problem though - the Internet connection is a bit flaky! could that be a problem?)

Also I notice following in the logs every now and then, but I don't think this per se causes an error.

InProcessAttachmentReader:readFailed:reason=SDS is closed

Also - I have seen following errors always -

[ FAILED ] AlexaDirectiveSequencerLibraryTest.sendDirectiveWithDifferentDialogRequestID
[ FAILED ] AlexaDirectiveSequencerLibraryTest.sendDirectiveWithoutADialogRequestID
[ FAILED ] AlexaDirectiveSequencerLibraryTest.multiturnScenario

Thanks

-abhijit

from avs-device-sdk.

Hi @gabhijit-shorelineiot

InProcessAttachmentReader:readFailed:reason=SDS is closed isn't an error as we explicitly put the SDS in conditions where this will happen during the test.

Would you mind attaching the full logs for those failed tests so we can better understand what is happening. A flaky internet connection might be the issue if it greatly causes the timings to be off.

Thanks,
Jamie

from avs-device-sdk.

@JamieMeyers

Attached is the file containing logs.

Thanks
-abhijit

err.txt

from avs-device-sdk.

Hi @gabhijit-shorelineiot,

This is a similar issue to #16, where your account is outside the US and we're not taking into account the incoming SetEndpoint directive as part of the test.

We're working on fixing this in an upcoming release. Until then, please use the following workaround:

Change AlexaDirectiveSequencerLibraryTest line 416 to read:

m_messageRouter = std::make_shared<HTTP2MessageRouter>(m_authDelegate, m_attachmentManager, "https://avs-alexa-eu.amazon.com");

Thanks,
Jamie

from avs-device-sdk.

Hi @gabhijit-shorelineiot,

Please make sure to redact any sensitive information you post in the forums. I've gone ahead and redacted your error.txt file to ensure your information is hidden. We're pushing a change in an upcoming release to ensure we don't log this privacy information in the future.

Thanks,
Jamie

from avs-device-sdk.

@JamieMeyers

Sure, I will redact going forward the personal information. Also I will let you know after making the change you suggested and running the code.

Thanks

-abhijit

from avs-device-sdk.

@JamieMeyers

I have managed to run all Integration tests so far. Only one Integration test is failing (For AudioInputProcessor.cpp). See the attached log.

This file is generated using latest SDK (0.4.1).

Thanks
err-201706121243.txt

from avs-device-sdk.

Hi Abhijit,

It sounds like you have the SDK up and working well. The error you are seeing occurs intermittently and we will be addressing it in a future release. The test was written to expect that no additional speak directive would be issued by AVS, but sometimes an additional speak directive is sent. This is not actually a failure in the sdk, but a difference in behavior by AVS from what was anticipated when this test was written.

Ken

from avs-device-sdk.

Hi Ken,

Thanks

As a next step I have started working on a simple App using libsound for interfacing with microphone. One thing that I observed, if one wants to build out of this source to use the AVS libraries, it's a bit challenging, what is missing is install() directives in CMakeLists.txt - I have added those and now I can build an app outside the source (it's very basic yet).

Should I create a separate issue for this and send a PR for that?

-abhijit

from avs-device-sdk.

Hi Abhijit,

That is definitely a gap in the current release, and we plan to address it soon. There's probably no need for a PR unless you have some specific concerns about how the SDK gets installed. We have done some preliminary work on this already with install directives for the headers and libraries, and will be including it in a future release.

Ken

from avs-device-sdk.

Hi Ken,

Thanks for the reply.

Also, what do you think would be the best way to get basic app reviewed, so that I know I am following the right path. What I am doing right now is -

1. Authenticate
2. Interface microphone with AudioInputProcessor and send a basic message (say "tell me a joke") and receive a response from AVS with SPEAK directive.

After this I will interface with the mediaplayer.

Thanks

-abhijit

from avs-device-sdk.

Hi Abhijit,

It sounds like you are on the right path. MediaPlayer integration should give you the end-to-end system where you can hear the responses and start to see a fully functioning system.

Ken

from avs-device-sdk.

Closing this out for now - please re-open if you have further questions on this topic.

Ken

from avs-device-sdk.