eac3to - audio conversion tool

[scarbrtj]

Originally Posted by ACrowley

@scarbrtj

I would not recommend SonicDecoder for EAC3 decoing!
You get full DRC+DialNorm +gain = very bad!

My problem is I could not get Nero to give me an AC3 file that sounded right. It skipped all over and I could not fix that. On the other hand, despite the issues you guys point out with using Sonic to decode DD+ --> AC3, the AC3 file sounds very good using an external high-end decoder played through big speakers!

Hey... what do you need to decode the .wav files eac3to generates if decoding to WAV? These do not play for me in WMP so that I can test them/listen to them.

[Penecho]

Originally Posted by madshi

The "---" should reach the end of the line (all but the last column) and then conversion should be done. There's no output file at all? Are you sure that the Nero decoder is correctly installed?

It works with the sonic decoder...

So i guess you were right! How can i reinstall the nero decoder?

Edit: i reinstalled nero 7 premium, but it did not help..
I am also noticed, that there is no "writing outoputfile" when i use nero, with sonic, there is one....



Cu

[SvT]

Originally Posted by Penecho

How can i reinstall the nero decoder?

Cu

You need to buy the HD plugin serial seperately from Nero. You put in your serial at "ProductSetup/Licence/Add" and you're good to go !

No SPAM just for help:
http://www.nero.com/eng/bluray-hddvd-video-plugin.html

Goodluck.

[Penecho]

I think i had that already, coz if not it would look like this (i uninstalled my nero and tested):


E-AC3, 5.1 channels, 1:49:16, 1536kbit/s, 48khz, dialnorm: -27dB
Removing dialog normalization...
Decoding with DirectShow (Nero Audio Decoder 2)...
Getting "Nero Audio Decoder 2" instance failed.


Cu


Penecho

[SvT]

Originally Posted by Penecho

Decoding with DirectShow (Nero Audio Decoder 2)...
Getting "Nero Audio Decoder 2" instance failed.

Cu

Just look at ProductSetup/Licence if your plugin serial is there ! (Not just the Nero serial). If so you're installed.

(I think your error shows if the plugin isn't properly registered).

[The_Keymaker]

Fellow Forum members,

The latest version (v1.49) of EAC3toGUI can be found here:

http://www.sendspace.com/file/xilu5v

Changes and features in this version include:

- Removed comma in audio delay output for values over 999ms.
- Added "Other Options" tab to allow custom or non-integrated options.
- Added *.dts as valid extension in the destination file.
- Changed command line output box to READ only.

This is an interim revision and I hope to release a version with all current options integrated.

As usual, remember to use the settings menu option to tell EAC3toGUI where the eac3to executable
is located.

Please report any problems or feature requests.

Regards,
The_Keymaker

[Thunderbolt8]

can anyone explain what the current status with xport 1.00 and seamless branching movies regarding audio sync is? link with info here: http://forum.doom9.org/showthread.ph...63#post1069463

I dont understand what we have to do now, or if we have to do something other than just demuxing with xport to keep audio in sync. will any joined part with LPCM audio now have +5ms delay and we have to fix that manually, either with altered fps rate or with that 'tail' procedure, that well have to cut off the endings and adjust both audio and video pts together at the end? or was this only an info he posted there and xport takes care of all that automatically?

[Nestorix]

Sorry to say, it is not working, tried all but cannot convert, not even automatic without adjusting something.....

[drmpeg]

Originally Posted by Thunderbolt8

can anyone explain what the current status with xport 1.00 and seamless branching movies regarding audio sync is? link with info here: http://forum.doom9.org/showthread.ph...63#post1069463

I dont understand what we have to do now, or if we have to do something other than just demuxing with xport to keep audio in sync. will any joined part with LPCM audio now have +5ms delay and we have to fix that manually, either with altered fps rate or with that 'tail' procedure, that well have to cut off the endings and adjust both audio and video pts together at the end? or was this only an info he posted there and xport takes care of all that automatically?

xport doesn't do anything automatically. Chopping off audio samples would be much too heavy handed (and annoyingly audible).

I'm not sure what the best remedy is. If you're re-encoding, then an extra frame (strategically placed) when necessary would be the best solution. If you're just re-muxing, why?

There is one positive note. The PES granularity of Dolby Lossless is 0.83 milliseconds. So for movies with Dolby Lossless audio, the A/V sync should be very close over many segments.

Ron

[Thunderbolt8]

Im remuxing, because I want the best possible quality at smallest possible size :P

which track count as dolby lossless, only TrueHD tracks, right? 0.83 is like nothing then. is this the same for DTS-HD (MA) tracks?? would it be possible for you to make a quick list of all tracks granularity for each .m2ts file, this should be helpful then when having movies with many parts to figure out how the delay has to be at the end. so for LPCM it was 5ms and for DD+ something like 32ms? trueHD 0.83 then? if you still need info about other audio tracks, then tell me, maybe I will be able to cut more samples, just to complete this.

LPCM: 5ms
DD+: 32ms
TrueHD: 0.83ms
DTS-HD MA: ?
DTS-HD HiRes: ? (same as MA?)

(I sent you both samples, MA with close encounters and DTS-HD HiRes with basic instict)


just assuming it is also 0.83 for DTS-HD MA for example, so when I remuxed close encounters with dts-hd converted to flac and this movie consists of 30 pieces, then the delay at the end would be ~25ms for the last part of the movie?
and in case of ratatouille with 31 pieces of LPCM the delay at the end would then be 155ms? and a 2 piece .m2ts movie with lpcm only ending up with 5ms, so this would be not worth correcting at all




btw. you mean positive delay, right (meaning we would have to slow down the fps rate of the video a little to make the video stream gradually longer) ?

btw² since you mentioned PES, do we have to use the -z option in xport now additionally for such files? (=e.g. xport -hz bla.m2ts 1 1 1) or still only -h ?

[Chumbo]

Originally Posted by Thunderbolt8

Im remuxing, because I want the best possible quality at smallest possible size :P

which track count as dolby lossless, only TrueHD tracks, right? 0.83 is like nothing then. is this the same for DTS-HD (MA) tracks?? would it be possible for you to make a quick list of all tracks granularity for each .m2ts file, this should be helpful then when having movies with many parts to figure out how the delay has to be at the end. so for LPCM it was 5ms and for DD+ something like 32ms? trueHD 0.83 then? if you still need info about other audio tracks, then tell me, maybe I will be able to cut more samples, just to complete this.

LPCM: 5ms
DD+: 32ms
TrueHD: 0.83ms
DTS-HD MA: ?
DTS-HD HiRes: ? (same as MA?)

(I sent you both samples, MA with close encounters and DTS-HD HiRes with basic instict)


just assuming it is also 0.83 for DTS-HD MA for example, so when I remuxed close encounters with dts-hd converted to flac and this movie consists of 30 pieces, then the delay at the end would be ~25ms for the last part of the movie?
and in case of ratatouille with 31 pieces of LPCM the delay at the end would then be 155ms? and a 2 piece .m2ts movie with lpcm only ending up with 5ms, so this would be not worth correcting at all




btw. you mean positive delay, right (meaning we would have to slow down the fps rate of the video a little to make the video stream gradually longer) ?

Humans don't notice a sync issue until the audio/video are off by 70ms or greater. Anything less is a non-issue, so I wouldn't waste so much time on it otherwise.

The real problem is when the small amounts per segment amount to a total offset that's greater than 70ms.

[Thunderbolt8]

Originally Posted by Chumbo

Humans don't notice a sync issue until the audio/video are off by 70ms or greater. Anything less is a non-issue, so I wouldn't waste so much time on it otherwise.

The real problem is when the small amounts per segment amount to a total offset that's greater than 70ms.

the problem is when knowing theres a little delay, even when it would possibly not be perceivable otherwise. it would just make me myself mad, I would look at the sync all the time and this would spoil the fun for me :S So I'll have to correct :S

[drmpeg]

LPCM: 5ms
DD+: 32ms
TrueHD: 0.83ms
DTS-HD MA: 10.666ms
DTS-HD HiRes: 10.666ms

Note that these are the maximum amounts of extra audio. On average, it will be half of the maximums. And yes, there's too much audio, so the video needs to slow down.

Don't use the -z option. That's for demuxing to PES packets, which almost no tool uses. Also, I think it's a little buggy (and never fixed since nobody uses it).

Here's an article on lip-sync tolerance.

http://www.tvtechnology.com/features...g_it_all.shtml

"The International Telecommunications Union (ITU) released a specification called BT.1359-1 in 1998. It was based on research that showed the reliable detection of A/V sync errors was between 45 msec audio leads video to 125 msec audio lags video. Remember, this is just the detectability region; the acceptability region is an even wider +90 to -185 msec."

Ron

[Thunderbolt8]

thanks for completing the list

hm the difference between max and average shouldnt make too much difference, at least not for trueHD and LPCM. but with movies with ~30 .m2ts files then it can already make a little difference (~50ms) for dts-hd and surely will in case of DD+ (~500ms). guess theres some experimenting needed :S

[Snowknight26]

When I join 3 m2ts files and demux the LPCM stream with xport, eac3to comes up saying that the format isn't recognized, though a graphedit graph of File -> ffdshow audio decoder shows that it is indeed (L)PCM.
Need a sample?

[shambles]

Originally Posted by drmpeg

Chopping off audio samples would be much too heavy handed (and annoyingly audible).

how would it be audible? if you calculate the amount of samples you need to chop off at the end of each audio segment, wouldn't you just be restoring it to its original length so it matches perfectly with the video?

for 48khz there's 48 samples per ms so you can get pretty damned accurate if you just have the patience to do it right.

(from earlier in this thread)

Originally Posted by drmpeg

The last audio PTS is 2752710 and the last video PTS is 2749008. The video ends at 2749008 + 3754 = 2752762. The audio ends at 2752710 + 450 = 2753160. 2753160 - 2752762 = 398. 398 / 90000 = 4.42 milliseconds too much audio.

from your example, take the 398 / 90000, multiply by 1000 = 4.4222..., multiply by 48 (48 samples/ms), then multiply by 18 (18 bytes/sample for 24bit/6channels) = 3820.8. but of course you can't chop off 3820.8 bytes off a file, and what we need to chop off needs to be dividable by 18 for 24bit/6ch files to keep the channel order correct, so divide it by 18 again = 212.2666..., then multiply 212 by 18 = 3816 which is the correct amount of bytes to chop off from the end of the pcm file. and if you want to be extra anal about it, like i do, you can subtract the 3816 from the 3820.8 and see that 4.8 bytes would count toward the next pcm file.

it takes a lot of time but i feel it's worth it. what would make it a bit easier would be if xport could be updated with a switch that displays only the last video and audio pts

[madshi]

Originally Posted by Snowknight26

When I join 3 m2ts files and demux the LPCM stream with xport, eac3to comes up saying that the format isn't recognized, though a graphedit graph of File -> ffdshow audio decoder shows that it is indeed (L)PCM.
Need a sample?

Yes, a sample would be helpful. But before uploading the sample please check whether you can reproduce the problem with the sample. If eac3to works fine with the sample, the sample won't help.

[madshi]

Originally Posted by drmpeg

Chopping off audio samples would be much too heavy handed (and annoyingly audible).

I'm wondering what the studio muxing software does when creating the multiple TS parts. I mean the studio master is one big video and audio file. Now if we join the TS parts and demux audio and video, audio is probably a few milliseconds too long. So obviously the TS muxing software must have added some audio data somewhere, right? How did it do that? Isn't it the most probably thing that the audio data is ever so slightly overlapping at the end of the first TS part and at the beginning of the 2nd TS part? So if you just join the PCM samples, a few milliseconds worth of audio data are played twice? Wouldn't chopping off the audio samples probably result in that we get the original audio track again? I don't really know, I'm just guessing...

Obviously things are much more complicated with AC3 compared to LPCM because we can't just chop off half an AC3 frame. So for AC3 demuxing the only reasonable solution would be to keep track of the "too much audio milliseconds". Once the delay sums up to over the length of an AC3 frame, we could then chop off one full AC3 frame (or several, if necessary). This way we could keep delay under the length of one AC3 frame.

Thoughts?

[rickardk]

Can someone please help me with this I would be very very thankful.

Just bought Bourne Ultimatum and I would like to remux it into TS and put it on my server so I can have it playable in Vista Media Center.


The edition I bought have a TrueHD track. And as there are no decoders that works with vista media center I will (if possible) convert it to flac. I don't really understand about sync problems and how to fix them.

Also the pulldown (29.97fps issue) must be removed from the VC-1 stream before it's muxed into a ts.

I know in theory what needs to be done. But not how (tools and exec) to do it.

HD DVD disc -> Demux VC-1 es and TrueHD -> remove pulldown from VC-1 es -> a perfect convert TrueHD to FLAC -> sync everything 100% -> mux into ts

If someone could write a step-by-step guide to get a perfect in Vista Media Center playable TS with the same exact sq as the truehd track (no dynamic range compression and stuff like that). I will just need one (TrueHD->FLAC) audio stream. And also a perfect VC-1 es without stutter or speed up. It would be the best christmas gift ever!



Sorry If this post belongs in another thread...

[madshi]

Originally Posted by rickardk

Can someone please help me with this I would be very very thankful.

Just bought Bourne Ultimatum and I would like to remux it into TS and put it on my server so I can have it playable in Vista Media Center.

The edition I bought have a TrueHD track. And as there are no decoders that works with vista media center I will (if possible) convert it to flac. I don't really understand about sync problems and how to fix them.

Also the pulldown (29.97fps issue) must be removed from the VC-1 stream before it's muxed into a ts.

I know in theory what needs to be done. But not how (tools and exec) to do it.

HD DVD disc -> Demux VC-1 es and TrueHD -> remove pulldown from VC-1 es -> a perfect convert TrueHD to FLAC -> sync everything 100% -> mux into ts

If someone could write a step-by-step guide to get a perfect in Vista Media Center playable TS with the same exact sq as the truehd track (no dynamic range compression and stuff like that). I will just need one (TrueHD->FLAC) audio stream. And also a perfect VC-1 es without stutter or speed up. It would be the best christmas gift ever!

Sorry If this post belongs in another thread...

Discussion about how to remux video doesn't really belong here. This thread is only about audio processing. For TrueHD -> FLAC decoding just demux the TrueHD audio track with EvoDemux and then use eac3to to recode that to FLAC. You'll get perfect quality, as long as you use the Nero or libav decoder. EvoDemux will tell you the delay value you need to use for the TrueHD track. You can feed that delay value into eac3to.