Audio glitches, problem finally solved !

[Lucius Snow]

Hello all,

I get audio glitches when i read a DVD on some standalones players. The problems occur always at the same moments of the videos while the image is clean by the way (no pixelate or stuff). Strangely, it doesn't happen on every standalone players, neither on a computer. I noticed however that old players are more concerned (> 5 years old). That's the case of my Pioneer DVD-626D.

The audio track is in PCM 48 kHz - 16 bit - Stereo (1536 kbit/s). The video track (encoded by Procoder 2) is PAL - CBR - 7800 kbit/s.

From the tests i made, i concluded that :

- this is not the media because i burnt on genuine Taiyo Yuden TYG03 and other DVD-R medias which are all very reliable

- this is not the authoring software since i tried rendering with DVD architect and DVD lab

- this is not the burning software since i tried burning with DVD architect and Nero

- this is not the total bitrate that would be too high for the players since the video stream is 7800 kbit/s CBR (so both streams audio+video streams are under 9,8 Mbit/s)

Notice also that :

- the problem doesn't occur when there's no video stream muxed with the audio one. By the way, i tried different MPEG2 encoders - and even at lower bitrates - in the case that the video stream would cause problem.

- i couldn't test yet with another DVD burner but i doubt it'd be the guilty (i own a Nec ND-3540).

Of course, when i play the audio source which is a WAVE PCM file on my PC, there's no problem. Then i thought that maybe the checksum, the CRC or something would be guilty, so i transfered the film on a DV tape and captured that one to get back a WAVE PCM file. It was finally identical and that didn't solve the problem.

What do you think about it ? I'm gettings nuts with that. I think i've tried all tracks but there may be another test to do ... ?

Thank you very much.

[setarip_old]

Hi!Strangely, it doesn't happen on every standalone players, neither on a computer. I noticed however that old players are more concerned (> 5 years old). That's the case of my Pioneer DVD-626D.It would seem that you've already determined the source of your problem (Your older standalone player)...

You might want to run a DVD/CD cleaning disc on your older standalone. Maxell makes one that sells for approximately $10US...

[Lucius Snow]

Hi!It would seem that you've already determined the source of your problem (Your older standalone player)...

You might want to run a DVD/CD cleaning disc on your older standalone. Maxell makes one that sells for approximately $10US...
Like i said, that doesn't occur with my standalone player only. This is also the case with some others players (but not all of them).

[jshumate]

> - this is not the total bitrate that would be too high for the players since the video stream is 7800 kbit/s CBR (so both streams audio+video streams are under 9,8 Mbit/s)


You told us nothing about how you made this video. You should be aware that video encoded with TMPGenc or recorded by video capture cards often spikes much higher than the CBR bit rate you set it for. CCE is one of the few encoders I know of that actually respects the bit rates it is given.

Go to http://www.tecoltd.com and get the free Bit Rate Viewer program. Have it analyze your video and see if your video bit rate is spiking much higher than you think. It would not be unheard of for it to go as high as 9000 kbps with some encoder or capture card solutions, despite the fact that you set the bit rate at 7800. Ever capture card I've ever used actually has the video bit rate spike higher than whatever you set the CBR value for, with ATI being the worst offenders in my experience.

[Lucius Snow]

I've spent my whole day to do new tests and i've just found that the problem comes from the video bitrate. So you got it right :)

I tried before with lower bitrate but not low enough.

So i encode with Procoder 2.0 : Mastering quality - 7800 kbit/s - CBR. For a DV interlaced source, it gives much better results than CCE or Mainconcept. Here are the full settings :

http://poubellelqt.free.fr/cap/procoder.jpg

And here is what Bitrate viewer says :

Num. of picture read: 78000
Stream type: MPEG-2 MP@ML CBR
Resolution: 720*576
Aspect ratio: 4:3 Generic
Framerate: 25.00
Nom. bitrate: 7800000 Bit/Sec
VBV buffer size: 112
Constrained param. flag: No
Chroma format: 4:2:0
DCT precision: 10
Pic. structure: Frame
Field topfirst: No
DCT type: Frame
Quantscale: Nonlinear
Scan type: ZigZag
Frame type: Interlaced

Average : 7617 kbit/s
Maximum : 8935 kbit/s (that's too much !)

Since my audio track takes 1536 kbit/s, i shouldn't go up than 8000 kbit/s for video since we're close to the 9,8 Mbit/s.

I took two audio glitches and looked what Bitrate viewer indicated at their time. That's 7700 kbit/s and 8200 kbit/s.

I had to set Procoder in VBR (3000 min / 6000 average / 7800 max) to avoid the audio glitches. But now the quality isn't as good as before. Since my video takes 52 minutes only, i want the maximum bitrate possible and i think i don't need a VBR. But it seems that Procoder exceeds the set values, especially in CBR. Should i try CQ (constant quality) in Procoder instead ? If yes, with witch settings ? I used CCE before but i let it down for interlaced sources ...

Thank you very much.

EDIT : Some glitches are gone but not of all them with VBR (3000 min / 6000 average / 7800 max) :(

[Lucius Snow]

Okay, new tests done.

6 samples encoded at the sequence when the glitches occur :

Test 1 : CCE SP - CBR 7800 kbit/s
Test 2 : CCE SP - CBR 7500 kbit/s
Test 3 : CCE SP - CBR 7200 kbit/s
Test 4 : CCE SP - CBR 7000 kbit/s

Test 1' : Procoder - CBR 7800 kbit/s
Test 2' : Procoder - CBR 7500 kbit/s
Test 3' : Procoder - CBR 7200 kbit/s
Test 4' : Procoder - CBR 7000 kbit/s

The audio stream is the same used for all videos, a PCM - 48 kHz - 16 bit - Stereo sample.

All of them got the glitch ! It sounded different each time, more or less audible, but there everytime. Bitrate viewer indicated peaks bitrates below 8000 kbit/s for test 3 et 3', and below 7600 kbit/s for test 4 ... So what the hell is going wrong ?

7600 + 1536 = 9136 kbit/s. We're in the norm. And my standalone player is supposed to handle that easily (Pioneer DVD-626D). Same for the others standalone players on which i test.

Well, i'm lost. I want to keep the audio stream in PCM for quality reasons. And i don't want to make a 3000 kbit/s video bitrate to avoid those glitches, especially since my video takes only 52 min, so the DVD can be filled !

What would you recommand ?

Thanks.

[mpucoder]

Well, despite your not believing it is the multiplexer (authoring) you might want to try another such as DVDMaestro, Scenarist, or MuxMan.

[Lucius Snow]

Well, despite your not believing it is the multiplexer (authoring) you might want to try another such as DVDMaestro, Scenarist, or MuxMan.
I already tried DVD-Lab but i can have a try to others to be sure. And the problem is that my whole authoring is already done in DVD architect :/

[mpucoder]

I'm not suggesting moving the entire project, but for testing purposes just mux the video and audio. DVDMaestro and Scenarist have templates for a single title autoplay, or you can start playback manually. MuxMan will create a single title autoplay by just selecting a video and audio file then pressing "start"

[Lucius Snow]

I tried with Muxman : same problem. I've also encoded the samples with both CCE (at CBR 7800) and Procoder (at 7200 CBR) with closed GOP (it was configured as opened in my first tests - see the screenshot above). Procoder sample failed at playback, CCE one succeeded *but* failed when i played again the sample after going back with the "search" button of the remote control.

I'm pretty sure the problem is linked to the video stream because the sound is perfect when the audio stream is played alone on the DVD.

I put online the sample i use for my test on which there're two glitches. This is an AVI - DV file (PAL - 720x576 - 25i) and a WAVE - PCM flle (48 kHz - 16 bit - Stereo). You can download them there :

http://poubellelqt.free.fr/cap/Render.rar

Of course, you may not see the problem if your standalone player isn't concerned. On four players tested, the glitches occured on two of them only.

I have to give the DVD next week to a compagny that'll make a glassmaster and 1000 copies. My DVD has to be 100% compliant because it'll be sold to public. The authoring is ready but i'm blocked here with that bullshit.

Pannic. :(

[kumi]

From the tests i made, i concluded that :

- this is not the media because i burnt on genuine Taiyo Yuden TYG03 and other DVD-R medias which are all very reliableI highly recommend you burn some tests on some quality TY DVD+R, with the booktype set to DVD-ROM (ImgBurn can probably set your drive's DVD+R booktype, and it's a top-notch burning engine, to boot.) Perhaps your players have a problem with the DVD-R media reflectivity/linking precision/ECC/itself. DVD+R (in my experience) has better compatibility with older standalones. Remember to set that booktype.

[Lucius Snow]

I don't have any DVD+R. I tried however to change the bitsettings to DVD-ROM with ImgBurn and DVDinfopro, both can't do it. ImgBurn "fails" and DVDinfopro indicates that i can only do it for DVD+DL :

http://poubellelqt.free.fr/cap/dvdinfopro.jpg

But if the problem comes from that media compatibility, why do the glitches always occur at the same times of the video ? Even when i make a total different authoring, with just a few samples tests and not the whole video, the glitches are there at these times. I mean that they never happen in random !

[mpucoder]

Is it possible that the audio itself has a problem? Have you played it alone either as a wav or in a DVD slideshow?

[kumi]

@Lucius Snow

Check your model here (http://club.cdfreaks.com/showthread.php?t=96140) for an updated firmware that allows DVD+R bitsetting. Hopefully it's just a limitation of the stock firmware.

[Lucius Snow]

Is it possible that the audio itself has a problem? Have you played it alone either as a wav or in a DVD slideshow?
Yes i have and it plays perfectly. I did another test : i desynchronised the audio and the video by cutting / pasting the audio sequence with the glitches to place them at another point of the video. Then the glitches moved to another moment ! So it has to do with the video stream which is working at the same time. I also noticed that the glitches occur especially when the screen fades to black (i do CBR). But when i encode to VBR with a low minimum bitrate, some glitches disappear (but not all of them).

@kumi : what can i do with that ?

http://club.cdfreaks.com/showthread.php?p=1122620#post1122620

EDIT : Just found the right link with explaination.

Thanks guys.

[Trahald]

Also suggestable is finding the bitrate that does not cause a peak over max (ie keep trying bitrates until you stop getting the glitch). the newer players have huge buffers and forgive a little peaking. once you find that bitrate.. encode only the segments with the chopiness issues at that lower bitrate and rejoin your mpeg .

of course if you try very low bitrates to test but are still unable to stop the glitching then there is a possibility its the audio and the other dvd players are forgiving the issue.

dvdlab comes with a decent bitrate viewer btw. 'Bitrate viewer' is not generally regaurded as being very helpful.

[Lucius Snow]

Okay, i have to go down to 2500 kbit/s in CBR to avoid the glitches completely. The video quality is awful then. It'd shock if i encoded the rest in 7500 kbit/s and join those MPEG in 2500 kbit/s. That'd be a solution but not great for quality.

By the way, i tested my burnt TYG03 DVD-R with Nero tool and here are the results :

http://poubellelqt.free.fr/cap/Nero.jpg

How do you interpret this ? Is it good ?

I've never used that tool before ...

@ kumi : i'm not really reassured with these non-official firmwares. From what i read on the posts, the results are mixed. That DVD i'll burn will be used to create a Glassmaster in order to press 1000 copies. From what i described, and the fact the glitches always occur at the same sequences of the video, not in random, and whatever the contents of the DVD (whole video or sample only), are you sure the media compatibility is the issue ?

[smok3]

a. how about doing a new 'original' like dumping it via vdub directstream copy or maybe even doing an huffyuv version of it before encoding... (well, maybe not..)

b. maybe try some other encoder like hc enc.

[Lucius Snow]

New tests done once again, and an interesting discovery !

I gave to DVD architect the original AVI-DV file, not encoded in MPEG. So it did itself the conversion on the fly. There's nothing to configure really except the bitrate (we don't even know if it's an average VBR or a CBR). I set it to 7000 kbit/s. A WAVE - PCM file was joined to it. That one was in 24 bit (i tried another resolution just to test but it had no incident - I used 16 bit before that test).

Into this authoring, i inserted also the MPEG created by Mainconcept (though Vegas) and Procoder from the same AVI - DV source. They are both set in 7000 kbit/s CBR as well.

Results : The glitches occured except on the sample encoded on the fly by DVD architect !

So i took the 3 VOB created in the rendering and opened them with DVD lab to get more details about their structure. I exported the reports into the .desc file that DVD lab propose (you can just open it with a text editor - that works too). So of course there're differences between the 3 samples and i'm looking for what settings i should change in Procoder to get the same as the one created successfully by DVD architect (which isn't very good quality by the way - but it is Mainconcept codec, like Vegas). The files are attached into a ZIP archive.

If some MPEG experts around can have a look and translate me the thing ... i'd be very happy :)

Thank you.

[Lucius Snow]

I highly recommend you burn some tests on some quality TY DVD+R, with the booktype set to DVD-ROM (ImgBurn can probably set your drive's DVD+R booktype, and it's a top-notch burning engine, to boot.) Perhaps your players have a problem with the DVD-R media reflectivity/linking precision/ECC/itself. DVD+R (in my experience) has better compatibility with older standalones. Remember to set that booktype.
I finally flashed my drive and bought DVD+R media (TDK) . So i set the booktype to DVD-ROM but the result is exactly the same : always glitches at the same parts of the film.

*desperated*

[Lucius Snow]

a. how about doing a new 'original' like dumping it via vdub directstream copy or maybe even doing an huffyuv version of it before encoding... (well, maybe not..)

b. maybe try some other encoder like hc enc.
I did a new original, AVI uncompressed, and another one, AVI DV (Canopus codec). Same problem. By the wat Huffyuv makes crashe the encoder.

I can't use HC since it seems to require AVIsynth that i don't know at all. Anyway, Canopus, CCE SP, Mainconcept and TMPGenc give all the same results.

Everything should demonstrate that the standalone player on which i'm doing the test has a problem. But i also heard the same glitches on another player.

After all those tests, i have noticed two weird details :
- the glitches always occur when the video gets black (fade or cut) just before the image's back
- the glitches disappear when DVD architect does the MPEG conversion itself on the fly (bitrate average : 7000 kbit/s).

[Lucius Snow]

Problem solved !!

By unchecking "Use strict GOP bitrate control" in Procoder and using VBR mode instead of CBR, the glitches disappear completely from the whole video. That option is checked by default and recommanded everywhere. I hope that disabling it won't make new problems. I haven't noticed anything going wrong on playback. Plus the quality is as good as before. I don't know the equivalent of this option in the other encoders i tried, so those still make making the glitches.

I must test on others standalones players to be sure it's resolved but it is already a success on my Pioneer DV-626D.

[Trahald]

Lucius

hehe.. you know i read something many eons ago.. (so might be wrong if my memory fails me) but i recall the advanced buffering mechanism (well.. advanced compared to mpeg1) of mpeg2 only kicks in for vbr encodes. i beleive the improvement you saw was strictly due to switching to vbr. welp.. glad you solved that one.

[Lucius Snow]

Lucius

hehe.. you know i read something many eons ago.. (so might be wrong if my memory fails me) but i recall the advanced buffering mechanism (well.. advanced compared to mpeg1) of mpeg2 only kicks in for vbr encodes. i beleive the improvement you saw was strictly due to switching to vbr. welp.. glad you solved that one.
This is also due to the option "Use strict GOP bitrate control". Switching to VBR only didn't solve the problem, whatever the encoders i used.

Do you think this option is absolutely necessary to make 100% compliant DVD ? It solved the problem but i'd like to be sure it didn't make new ones ;)

[Trahald]

is strict gop control needed (makes sure bitrate on a gop level doesnt drop below minimum requested)? dunno. i have never had an issue with setting bitrate minumum to 0. i didnt see anything in dvd demystified mentioning a minimum nor through doing some quick googling.. maybe mpucoder may want to chime in on this one tho. imo if it passes your picky machines it should do the trick.

[sherman_oakes]

Hi there lucius, i'm having sync problems, also with dvd architect. I'd have imagined setting strict gop bitrate control would've been a good thing.

[Audionut]

http://poubellelqt.free.fr/cap/Nero.jpg

How do you interpret this ? Is it good ?



That is VERY bad, and is most likely the cause of your problems, regardless of what else you have done that has fixed the problem.

http://club.cdfreaks.com/showthread.php?t=80545

should not exceed 280 PI errors

According to our test the specified max PI sum of 280 for good discs seems to be a good guideline, as a few readers have problems reading discs when the PI failures is over 300 and many players starts to have problems when the PI failures reaches 600 and most players have problems with discs exceeding 900 PI errors or more.

edit: and as you can see from your scan, you have errors larger than 1400.

Here's a scan for comparision.

http://img148.imageshack.us/img148/4949/compare4tp.png

[Lucius Snow]

Thanks for the feedback Audionut.

I gave the DVD-R to the factory which pressed it. So here is the result of the pressed one now. It looks much better, does it ?

http://poubellelqt.free.fr/cap/_NEC____DVD_RW_ND-3540A_1.W9_21-May-2006_14_42.png

[sherman_oakes]

If the DVD-R shows such great change when eventually pressed, would it be better simply to give the original high bitrate version?

[Lucius Snow]

Well the glitches didn't come from the media since i tested it with others medias such as Verbatim and TDK. Anyway the DVD are already pressed, i can't go back. And the result is simply perfect.

[sherman_oakes]

Hi there lucius so what are the vbr settings used on the DVD-R? What does the factory do to the content or is it just the burning process that ensures everything works out fine? In which case sending a high bitrate CBR would've been better, even though it glitches on some players. Also would it have been better to just send an image of the DVD rather than a DVD, since then the media quality is not a problem? Do they handle it that way?

[Lucius Snow]

The factory didn't change anything. The problem has been solved by changing disabling "Use strict GOP bitrate control" and switching to VBR. So it was already fine when i shipped them the DVD-R.

[Audionut]

It looks much better, does it ?

Yes, indeed it looks much better than your original scan.

However:) ,

Either your drive is having problems reading the disc's, or most likely, whoever is making those disc's needs to quit there day job. Seriously. For a pressed disc!!!

For a 1x read scan, that result is not very good. Given a bit of time and normal wear and tear, that disc would quickly become unreadable.