EXPERIMENTAL: New MPEG-TS Rewriter for HLS (TVE)

[speedingcheetah]

I had this option enabled on my other server, that my mother uses.
I forgot i had enabled it.
She was away for some time, now is back full time.

It was causing Pluto channels, ie Judge Judy, to have excessive buffering/pausing every so often. Have to close out the channel and reopen.
I disabled it, and never had the issue since.
This was a few weeks ago.
Has the rewriter been updated since?

Is it possible to make this option selectable per channel? Since it is only needed, it seems, on Discovery based channels.

[tmm1]

Yes there have been several fixes recently. We are aiming to make the new rewriter the default for everyone soon. If you experience issues let us know.

[mnwxman132]

Tests went great last night on discovery and history channels prime time lineup.

Programs had full length and i didnt notice any dropped frames.

[eric]

We are continuing to have confidence that the new rewriter is handling all of the cases we have looked at better than the old one. Please let us know if you run into any issues with the new one that we need to look into further.

[speedingcheetah]

It's been working fine so far.

[Absenm]

When @tmm1 mentioned that you would soon be making it the default I started using it. I have noticed no issues to report. It’s working great for me on my admittedly generic Synology/YoutubeTv setup. Hopefully once this becomes a default maybe it will make things easier with the CC for Pluto dream.

[chDVRuser]

After enabling this, I'm no longer getting the MTS stats in the log for recordings, like this

Issues recording TVE from TLC, Discovery, TRVL and Food

I was about to post that I don't have these issues. I found a couple recent Magnolia channel recordings where the 35 minute recordings show as 15 minutes and the timestamps are messed up. To fix them, I remuxed them with ffmpeg. Just enabled the experimental rewriter. I see a dead giveaway in the log where the [MTS] highest_pts seen is not high enough. Should be >= 2100 for a 35 minute recording. highest_pts=1247.165700 (20 minutes, 47 seconds) [MTS] Statistics for "TV/Magnoli…

Can they be added back.
highest_pts, saw_pcr, skipped and unhandled_packets are useful.

[eric]

Not really as such. The new rewriter doesn't operate the same way as the old one, so skipped= and unhandled_packets= aren't a thing that is meaningful in the new system.

I can look at adding back highest_pts=, but the meaning in the new system is somewhat different from what it was before...

[chDVRuser]

@eric

I can look at adding back highest_pts=

Thanks. Mainly useful for interrupted recordings so I can quickly locate the spot in my video editor where Channels DVR stitched the stream back together.

saw_pcr=false meant something was changing the PMT to say the PCR was in a PID when there was actually no PCR in the stream EXPERIMENTAL: New MPEG-TS Rewriter for HLS (TVE) - #84 by chDVRuser

[chDVRuser]

I enabled this 2 weeks ago on all four DVR servers and haven't seen any issues with recordings or Live.
Using TVE and M3U sources.

[chDVRuser]

Will post if any issues with these 21 one hour recordings this week (Shark Week on Discovery).

Day one 4 hrs of back-back recordings, all good
Day two 3 hrs of back-back recordings, all good
Day three 3 hrs of back-back recordings, all good
Day four 3 hrs of back-back recordings, all good
Day five 3 hrs of back-back recordings, all good

And after recording, updated to

DVR Pre-Release Notes

v2023.07.27.2357 • FIXED: Experimental MPEG-TS rewriter could run into problems on CNN TVE feed causing glitching video

and

DVR Pre-Release Notes

v2023.07.28.1829 • FIXED: Video glitches that could happen after the last fix when using the Experimental MPEG-TS rewriter

just to make sure I'm testing w/latest version

Day six 3 hrs of back-back recordings, all good
Day seven 2 hrs of back-back recordings, all good

All 21 hours without any issues.

Pausing the pass until next year.
Thanks @eric for the work on the new rewriter!

[Ahwman]

After upgrading to v2023.07.27.2357 with the MPEG-TS Rewriter enabled, Newsmax (channel 6764) is now unwatchable as it stutters terribly. It was smooth as butter in the prior beta version…

[eric]

@fancybot

v2023.07.28.1829

• FIXED: Video glitches that could happen after the last fix when using the Experimental MPEG-TS rewriter

This latest update should resolve the issues you saw.

[Ahwman]

That did the trick! Thanks so much Eric. Have a blessed weekend my friend…

[eric]

After investigating further I was able to figure out how wire up the rewriter statistics for the new rewriter. Some of the values will always be 0 with the new system (skipped, unhandled_packets, invalid_pts, invalid_dts), but the other ones that sounded like the most important ones will be working.

@fancybot

v2023.07.30.2322

• IMPROVED: Provide rewriter statistics for Experimental MPEG-TS rewriter
• NEW: Log rewriter statistics after a completed Live TV stream

[chDVRuser]

Thank eric, looks great

2023/07/30 17:00:29.526430 [MTS] Statistics for "TV/Fixer Upper/Fixer Upper S05E18 2018-04-03 Rock Star Renovation 2023-07-30-1646.mpg": skipped=0 unhandled_packets=0 discontinuity_detected=15 transport_errors=0 invalid_pts=0 invalid_dts=0 saw_pcr=false saw_pmt=true highest_pts=856.434744

highest_pts=856.434744 Exactly correct!

Even though it shows saw_pcr=false, the recording does have a PCR in the right place, the video PID. Not sure if you're rewriting the PCR?

[eric]

@chDVRuser

Not sure if you're rewriting the PCR?

Yes, we're adding a synthetic PCR if one hasn't been seen in 100ms to conform to the MPEG-TS spec. That saw_pcr=false is indicating that we did not see one in the original stream.

[eric]

@fancybot

v2023.08.05.1951

• NEW: Experimental MPEG-TS rewriter is now the default. The old one is available for debugging under Experimental settings.

The new MPEG-TS rewriter is now default for everyone. Thanks to everyone who helped with the testing!

[speedingcheetah]

“The old one….” Think is what u ment to say in the release notes is the one that’s available in experimental settings