search

cta-wave / Test-Content-Generation

Provides all information and scripts how the CTA Wave Test Content is generated
BSD 3-Clause "New" or "Revised" License
4 stars 6 forks source link

Absence of audio track prevents testing AV sync #14

Closed nicholas-fr closed 3 years ago

nicholas-fr commented 3 years ago

Test content is missing an audio track (cannot test AV sync). For video tests, propose to add a switching set with a single audio track with mezzanine audio encoded using AAC core media profile.

jpiesing commented 3 years ago

December 15th This is an issue with the script Rodolphe wrote so deferred until staffing issues are resolved.

rbouqueau commented 3 years ago

Could you please give us a pointer on the audio available content (or how to generate it) ? I don't see anything in http://dash.akamaized.net/WAVE/Mezzanine.

jpiesing commented 3 years ago

Could you please give us a pointer on the audio available content (or how to generate it) ? I don't see anything in http://dash.akamaized.net/WAVE/Mezzanine.

The .mp4 files have both video and audio in them.

rbouqueau commented 3 years ago

Excellent, then it indeed looks like a bug. I'm taking this issue.

jpiesing commented 3 years ago

Excellent, then it indeed looks like a bug. I'm taking this issue.

I know it's a little confusing - these files would ideally have been prores but ffmpeg prores output didn't seem to work properly. See https://github.com/cta-wave/mezzanine/issues/15

rbouqueau commented 3 years ago

What was the plan about audio for the single track switching sets? Is audio only for the ss X1?

jpiesing commented 3 years ago

What was the plan about audio for the single track switching sets? Is audio only for the ss X1?

We need content for both of the following; a) video-only switching set tests (8.5, 8.11) b) tests for CMAF Presentations that need both video and audio (9.x)

For the CMAF presentation test content, we can use the audio that's in the mezzanine content & just encode it as HE-AAC. This audio also has beeps added that should be synchronised with the flashes in the audio

@rbouqueau Does this answer the question or have I missed something?

rbouqueau commented 3 years ago

Thank you Jon. I think it is clear enough for now.

andyburras commented 3 years ago

I'm still a bit confused as to the intended make up of the 8.x test content. https://github.com/cta-wave/dpctf-tests/blob/master/sample.csv seems to imply that the 'avc' tests don't have audio (there is no audio MPD listed). However, there is an audio Representation within the "video" MPD, that the Test Runner does play. In the future, is this audio intended to be part of the test analysis?

jpiesing commented 3 years ago

Tests for 8.x don't need audio. If the MPD includes audio then this should be ignored by the HTML+JS code for that test. The observation framework should ignore it.

Jon

From: Andy Burras - EDT @.***> Sent: 10 May 2021 17:03 To: cta-wave/Test-Content-Generation Cc: Jon Piesing; State change Subject: Re: [cta-wave/Test-Content-Generation] Absence of audio track prevents testing AV sync (#14)

[External Email] This message was sent from outside of the company. Please do not click links or open attachments unless you recognize the source of this email and know the content is safe.

I'm still a bit confused as to the intended make up of the 8.x test content. https://github.com/cta-wave/dpctf-tests/blob/master/sample.csv seems to imply that the 'avc' tests don't have audio (there is no audio MPD listed). However, there is an audio Representation within the "video" MPD, that the Test Runner does play. In the future, is this audio intended to be part of the test analysis?

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHubhttps://github.com/cta-wave/Test-Content-Generation/issues/14#issuecomment-836810076, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ACAY6KEPCBYDP57Z6O4QX5TTM7YUPANCNFSM4USQKJYA.

jpiesing commented 3 years ago

I'm still a bit confused as to the intended make up of the 8.x test content. https://github.com/cta-wave/dpctf-tests/blob/master/sample.csv seems to imply that the 'avc' tests don't have audio (there is no audio MPD listed). However, there is an audio Representation within the "video" MPD, that the Test Runner does play.

For the 8.x tests for AVC (or indeed any video media profile), the audio in those MPDs could be ignored.

In the future, is this audio intended to be part of the test analysis?

Not for the 8.x tests.

jpiesing commented 3 years ago

@nicholas-fr As far as I can see, the content includes both video and audio Adaptation Sets. Do you agree? Can this issue be closed?

andyburras commented 3 years ago

I'm still unclear on the usage of this audio. It's not required for the 8.x tests. I thought for the 9.x Presentation playback tests, that Test Runner used one MPD for the video and a different MPD for the audio, or is that not the case?

jpiesing commented 3 years ago

I'm still unclear on the usage of this audio. It's not required for the 8.x tests. I thought for the 9.x Presentation playback tests, that Test Runner used one MPD for the video and a different MPD for the audio, or is that not the case?

For the 8.x tests, the audio can be ignored. For the 9.x tests, the same MPD can be used for both video & audio.

nicholas-fr commented 3 years ago

Resolved, with the caveat that ss1 cannot be used for AV sync testing as the same audio track is used for all the different video representations with different frame rates (not to imply that ss1 was intended for AV sync testing, just a clarification for the avoidance of doubt).