• AnyStream is having some DRM issues currently, Netflix is not available in HD for the time being.
    Situations like this will always happen with AnyStream: streaming providers are continuously improving their countermeasures while we try to catch up, it's an ongoing cat-and-mouse game. Please be patient and don't flood our support or forum with requests, we are working on it 24/7 to get it resolved. Thank you.

Am I the only one seeing audio sync problems with some filters

J

Jong

Well-Known Member
Thread Starter
Joined
Jan 6, 2008
Messages
1,601
Likes
0
Most of you no doubt will have seen my exchange with James in the support thread.

My results are rock solid and reproducible - with some filter combinations the audio sync varies far too much causing dropouts or, if you increase the latency to avoid them, very audible ("warbling" in music) and visible audio sync problems. With others all is perfect, but who knows what the next version of any filter will bring or what other codecs we will need in the future. As you will have gathered (!) it seems to me the buffering in Reclock is seriously broken at the moment - all is fine if it is not needed (well behaved filter chain) and things fall apart if it is! I suspect this may even have an impact on otherwise well behaved players/filters when Windows decides to go off and do something else for a few milliseconds and Reclock needs to compensate.

I'm surprised that no one else seems to have raised this as an issue, since nearly all of us are AV perfectionists! I cannot really blame James for not jumping to look into something only I am raising. Do you all think I am mad?!

I'm very interested in your thoughts!
 
I've started having no end of problems with Blu-ray Audio Sync using TMT http://forum.slysoft.com/showthread.php?t=26426 . If seems to me that if something causes an issue with the smooth playback of sound or video a cycle starts up where different processes are trying to resync the two but it never gets there.
 
I did think that might have been why AC3Filter seems to cause so much trouble here - it has its own built in AV sync correction. But I turned it off and things were just as bad! So either it is still messing around with the sync in some way or that is not it.

It is weird the way that MPC-HC with Cyberlink's MPEG-2 decoder settles to a reasonable (if not great) +/- 8ms or so, but with Nvidia's decoder it is swinging wildly by +/-250ms if I use a 1000ms buffer. And the same Cyberlink decoder in PowerDVD (OK, with quite a lot of other "special" Cyberlink filters thrown in) settles to +/-1ms most of the time +/-3ms almost all of the time.

It is not as simple as the Nvidia decoder messing up AV sync. If I use s/pdif passthough sync is +/-1ms and all is perfect. Something happens when some filters interact with the Reclock resampler.
 
Last edited:
jmone said:
I've started having no end of problems with Blu-ray Audio Sync using TMT http://forum.slysoft.com/showthread.php?t=26426 . If seems to me that if something causes an issue with the smooth playback of sound or video a cycle starts up where different processes are trying to resync the two but it never gets there.
Click to expand...

I told you, TMT .128 & .129 are broken, *DO NOT USE THEM*, they are out of sync "by design". Disable ReClock and try them with 24fps BD, or PAL DVDs. Audio & Video will drift *seconds* out of sync (not to mention stuttering, pixelation, and the usual crashes under XP which began with all versions after .113)
TMT is currently a complete mess. Unusable. At the moment I am using .120 under Vista, as this version doesn't downsample. ;)
If you use AnyDVD HD taking care of AACS & BD+, I don't see much reason to always use the "newest and greatest" version.
Especially if both Cyberlink & Arcsoft drive their products more and more downhill. :mad:
 
Jong said:
Most of you no doubt will have seen my exchange with James in the support thread.

My results are rock solid and reproducible - with some filter combinations the audio sync varies far too much causing dropouts or, if you increase the latency to avoid them, very audible ("warbling" in music) and visible audio sync problems. With others all is perfect, but who knows what the next version of any filter will bring or what other codecs we will need in the future. As you will have gathered (!) it seems to me the buffering in Reclock is seriously broken at the moment - all is fine if it is not needed (well behaved filter chain) and things fall apart if it is! I suspect this may even have an impact on otherwise well behaved players/filters when Windows decides to go off and do something else for a few milliseconds and Reclock needs to compensate.

I'm surprised that no one else seems to have raised this as an issue, since nearly all of us are AV perfectionists! I cannot really blame James for not jumping to look into something only I am raising. Do you all think I am mad?!

I'm very interested in your thoughts!
Click to expand...
One thing - the audio buffer is allocated by the sound card driver. The driver is queried for buffer positions where the next chunk shall be placed.
1s (with 6-8 channels, high sampling rates) is quite a lot, and I've seen some drivers (Auzentech) go crazy with buffer sizes > 0.8s

The default values are default values ... for a reason. Even if you have the possibility to change it, this doesn't say it'll work (like the "use hardware resampling" checkbox, which very, very often appears to work, but introduces distortion with many cards, at least under XP)
 
That would be fine if the defaults worked, which they don't with some filter combos. And it doesn't explain why MPC-HC with mkv (AC3 6-channel) and PowerDVD (DVDs and Blu-ray) can have very low audio sync variability, even with the same source and same buffer, yet some filters cause Reclock to go crazy.

Even if we could not fix it, if we could find the reason and understand it it would be something. At the moment it feels like a game of chance. :confused:
 
Jong said:
That would be fine if the defaults worked, which they don't with some filter combos. And it doesn't explain why MPC-HC with mkv (AC3 6-channel) and PowerDVD (DVDs and Blu-ray) can have very low audio sync variability, even with the same source and same buffer, yet some filters cause Reclock to go crazy.

Even if we could not fix it, if we could find the reason and understand it it would be something. At the moment it feels like a game of chance. :confused:
Click to expand...
Yes, it is quite .... puzzling. I'm still looking into this.
 
Thanks James. Let me know if I can help in any way.
 
This filter inconsistency is driving me up the wall.

For the last two days I have been struggling with occasional but very disruptive "WAVE Timeouts" with PowerDVD and MPC-HC playing DVDs. Both had the same problem. Regardless of buffer size (yes, i did try the default!!) I would get seemingly random timeouts. They could occur at any time but typically every 1500-3000 seconds. Certainly in a movie they would occur a few times. The rest of the time all is perfect, especially with PDVD - very very low audio "sync" figures - then, out of nowhere, a timeout! As you will all know I am sure, timeouts are MUCH worse than a drop/repeat. They result in a complete flush and re-fill of the buffer. In MPC-HC this results in the video freezing for the time it takes for this to happen (0.5-1sec). In PowerDVD (I think due to its own "timestretching filter", for DVD only) it goes crazy for about 10 seconds trying to compensate, with extreme chipmunk voices etc!

I have no idea what caused this to start happening. It may have always been there, as I used to use TheaterTek with ffdshow for DVDs before the other problems came up. But it may all have been introduced around 1.8.3.0-1.8.3.3, when my problems with TT/Nvidia video decoder showed up. It may be part of the same problem.

I have spent the best part of 3 days, 24 hours a day (literally!), running various tests (unattended most of the time, before you get worried for my mental health!) with various different players and filters. In addition I have also been inspecting event logs at the time of the timeouts, looking at files that have been written to at the time of the timeouts(!), killing various "nice to have" processes and services etc. etc. in case there was something rogue causing the problem. I also tried various Reclock settings, changing the resample quality, turning AC3 encoder on and off (which just replaced wave timeouts with, even worse, Directsound timeouts), even changing the resampler to/from Yesgrey's version and, I am sure, a few other things too. At the end of it all I could not shift the problem at all. All seemed perfect. Nothing strange was happening on the PC at the time of the timeouts, although of course Windows is always rotating around various services and processes. Playback was perfect, until these random very distracting timeouts. I was stumped.

More in desperation than anything I decided to re-introduce AC3Filter to the mix (in MPC-HC). I had always been using it to decode AC3 from TT (when things were working), but had taken it out when it seemed to make the Nvidia video decoder problem worse. As if by magic AC3Filter has made the problem go away. It is not even decoding AC3 for MPC-HC, MPC's own internal filter is doing that. It is just passing through PCM to Reclock. It is set to resample all audio to 48Khz (to fix the problem with apple trailers that use 44.1khz audio), but for AC3 on DVD the sound is already 48khz, so AC3Filter should be doing next to nothing. Amazingly this results in PERFECT playback in MPC-HC. I can enable and disable AC3Filter, play the same DVD for hours on end, and with AC3filter - no timeouts - without AC3filter - one timeout at least every 3500 seconds (on average every 1500 seconds or so). :confused::confused:

I have no idea what AC3Filter is doing. I guess it might be stabilizing something or maybe it is just the additional delay that AC3Filter adds to the audio stream?! It is not a magic fix, with my Nvidia video decoder problem it seemed to make things worse. It just seems that I now need a very carefully crafted filter chain to get Reclock to work as it should :(.

Sorry for the long rambling essay, but I can only repeat what I have said before - I think fairly recently something has happened in Reclock to break its buffering algorithm and make it very sensitive to the timing of the filter chain.
 
Last edited:
You must log in or register to reply here.
Back
Top