4.0.5 tries to use AAC renderer for AC3 after channel change

[sgilani]

You've got lots of fun logging hooks. Here you go.

[sub]

If you repeat that test, but leave it on the second channel for a minute or so, then copy the live tv buffer file (so it doesn't get deleted when you stop live tv), then play the file - does it have audio? is it AC3 or AAC? It looks like it's reading the descriptor from the device, and it's reporting AAC.

[sgilani]

The file doesn't play in NPVR or VLC, but plays fine in the Movies & TV player that comes with Win10 (and it has AC3 audio).

Here's a link to the ts file.
https://drive.google.com/file/d/1QeQvePs...sp=sharing

i have a hunch that while the channel changes, there's no audio signal coming from the STB and the HDPVR2 falls back to AAC. when the audio signal comes back as AC3, its too late because NPVR has already hooked up the playback graph and expects AAC.

[sub]

i have a hunch that while the channel changes, there's no audio signal coming from the STB and the HDPVR2 falls back to AAC. when the audio signal comes back as AC3, its too late because NPVR has already hooked up the playback graph and expects AAC.

Yeah, it is definitely something like this - I can see the AAC audio track in the file. The channel is initially starting out as AAC (and I very briefly see your set top box OSD graphics), then switching to AC3.

In your config.xml, do you have <WaitLongerBeforeRecordingCapture>true</WaitLongerBeforeRecordingCapture>? If not, try setting it.

[sgilani]

In your config.xml, do you have <WaitLongerBeforeRecordingCapture>true</WaitLongerBeforeRecordingCapture>? If not, try setting it.

That didn't change anything, so I made a batch script to sleep for 2 seconds after blasting, and that seems to work. What's the unit of measure for the delay timing to wait longer? It's set at 6500, and it certainly didn't feel like 6500ms.

[sgilani]

one last thing If I switch from a channel with AC3 to one wth AAC (from iptv), I get the same "you dont have an AAC decoder" message. Is the software trying to reuse the previous graph or something?

Regular and "temp" logs attached.

[mvallevand]

That didn't change anything, so I made a batch script to sleep for 2 seconds after blasting, and that seems to work. What's the unit of measure for the delay timing to wait longer? It's set at 6500, and it certainly didn't feel like 6500ms.

I remember when sub added this option and it never worked the way I hope with the pause after blasting so I had to add it in a batch file too.

Martin

[sub]

Looking at the code, it looks like the WaitLongerBeforeRecordingCapture / WaitLongerDelay logic only happens for the recordings. I can't remember the reason I might have explicitly excluded live tv.