eac3to - audio conversion tool

[madshi]

eac3to v2.48 released

http://madshi.net/eac3to.zip

v2.48
* 96kHz LPCM tracks in (M2)TS and EVO/VOB containers didn't work correctly
* "Applying (E-)AC3 delay" now only shows if the bitstream is actually modified
* fixed crash in MP2 reader when checking some PCM tracks
* added support for MLP formats 13 - 16
* improved/corrected MLP channel descriptions
* MLP parser sets correct channel mask
* added proper channel remaps for libav MLP decoding of "funny" channel formats
* added proper channel remaps for Nero MLP decoding of "funny" channel formats
* added proper channel remaps for Nero AC3 decoding of "funny" channel formats
* when doubling 7th channel the channel mask is set correctly now
* channel mask is corrected if a decoder doesn't output all channels
* channel mask is corrected if a channel downmixing is performed

I've checked out the ArcSoft Audio Decoder. It seems to work well enough - but only when using it in GraphEdit and when manually adjusting the parameters to the correct values. When accessing the decoder from within eac3to the ArcSoft Audio Decoder always outputs stereo, only. I've not found a (documented/official) way how eac3to could convince the decoder to output more than 2 channels, sadly...

[menlvd]

madshi
when rename w64 file 2 pcm file -> eac3to detecting file fine, but the channel masking is't right
original -> produced by eac3to
L -> SL
R -> SR
C -> L
LFE -> R
SL -> C
SR -> LFE
maybe U can look at this
simple need channel remaping -4,5,0,1,2,3

[rack04]

Originally Posted by rebkell

tsmuxer might possibly remove the pulldowns, I'm assuming hd-dvd, blu-ray doesn't normally have pulldowns in their h264 files.

Yes it is Transformers HD DVD.

[Thunderbolt8]

Originally Posted by madshi

eac3to v2.48 released

thanks!

regarding all those mlp changes/fixes, does that mean that some former converted flac tracks could possible be 'wrong' or not working correctly?

and regarding the problematic mpeg2 .ts files, is this already adressed here as well and I can try it again with 2.48?

[rebkell]

Originally Posted by rack04

Yes it is Transformers HD DVD.

madshi recommended h264info, and he would be a much better source than I ever hope to be on the subject. If you plan on re-encoding the h264, then DGAVCDec will recognize the pulldowns and frame serve them minus the pulldowns.

I've used tsmuxer in the past, but found that eac3to seems to always keep audio and video in sync and I never have any problems dealing with anything it ouputs.

I'm probably not the normal eac3to user, I normally just extract the video track and convert one of the English audio tracks to a lower bitrate ac3 stream.

[rack04]

Originally Posted by rebkell

madshi recommended h264info, and he would be a much better source than I ever hope to be on the subject. If you plan on re-encoding the h264, then DGAVCDec will recognize the pulldowns and frame serve them minus the pulldowns.

I've used tsmuxer in the past, but found that eac3to seems to always keep audio and video in sync and I never have any problems dealing with anything it ouputs.

I'm probably not the normal eac3to user, I normally just extract the video track and convert one of the English audio tracks to a lower bitrate ac3 stream.

Yes I'm going to re-encode. Using eac3to I change the container to mkv the pulldown is removed, i.e 23.976. When I extract the h264 and use DGAVCDec to index it reports 29.97 and audio video is out of sync. I'll try h264info and report back.

[rebkell]

Originally Posted by rack04

Yes I'm going to re-encode. Using eac3to I change the container to mkv the pulldown is removed, i.e 23.976. When I extract the h264 and use DGAVCDec to index it reports 29.97 and audio video is out of sync. I'll try h264info and report back.

Does it report 29.97 after you let it(DGAVCIndex) finish? It will report 29.97, but when it finishes it will say that it has pulldowns and would you like to remove them(or something similar to that), click yes and it will then serve them without the pulldowns, at least that was my experience, I think it was with Beowulf that I used it for.

[rack04]

Originally Posted by rebkell

Does it report 29.97 after you let it(DGAVCIndex) finish? It will report 29.97, but when it finishes it will say that it has pulldowns and would you like to remove them(or something similar to that), click yes and it will then serve them without the pulldowns, at least that was my experience, I think it was with Beowulf that I used it for.

Don't think I let it finish. I'll try again thanks.

EDIT1: I just re-ran DGAVCDec and it never asked to remove the pulldowns.

EDIT2: Change the setting to FORCE FILM and it appears to have worked. Thanks.

[rickardk]

Originally Posted by madshi

I think tebasuna51 can probably answer this best. However, I've just done some tests with funny AC3 channel formats and sometimes the Nero decoder boosted volume on some channels. But normally in my experience the decoders recommended to be used by eac3to (see first post of this thread) do not do any do anything "wrong". In other words: I tried encoding a WAV file in AC3 and DTS. And decoded them again. And encoded them again. And decoded them again. The volume of all channels were unchanged in my tests. I didn't test this with all decoders, but I think the decoders recommended for use with eac3to should be fine.

I asked because I read that some software decoders boost the lfe channel with +10dB (just as hardware ac3/dts decoders does in your regular av receiver).

Great to know that this is not the case with eac3to. Because then you can have your receiver or software like ffshow do the lfe +10dB boost on all remuxes created with eac3to.



This might be ot, but how do I use Graph Edit to decode my 7.1 DTS-HD MA tracks with Arcsoft TMT audio decoder?

I extracted the audio tracks long ago to .dts files. I can't connect the output pin from Sonic HD Demuxer to Arcsoft HD Audio decoder without having Sonic Cinemaster Audio decoder or ffdshow conncted in between.

[madshi]

Originally Posted by menlvd

when rename w64 file 2 pcm file -> eac3to detecting file fine, but the channel masking is't right
original -> produced by eac3to
L -> SL
R -> SR
C -> L
LFE -> R
SL -> C
SR -> LFE
maybe U can look at this
simple need channel remaping -4,5,0,1,2,3

w64 files are simply not supported yet.

Originally Posted by Thunderbolt8

regarding all those mlp changes/fixes, does that mean that some former converted flac tracks could possible be 'wrong' or not working correctly?

Blu-Rays and HD DVDs are TrueHD and *not* MLP. So for most people all the MLP related changes have no effect whatsoever. MLP is mainly used by DVD-Audio.

Originally Posted by Thunderbolt8

and regarding the problematic mpeg2 .ts files, is this already adressed here as well and I can try it again with 2.48?

Yes.

Originally Posted by rickardk

how do I use Graph Edit to decode my 7.1 DTS-HD MA tracks with Arcsoft TMT audio decoder?

I extracted the audio tracks long ago to .dts files. I can't connect the output pin from Sonic HD Demuxer to Arcsoft HD Audio decoder without having Sonic Cinemaster Audio decoder or ffdshow conncted in between.

Don't know if you can directly decode demuxed DTS files in GraphEdit with ArcSoft. Maybe it's possible somehow, not sure.

In the meanwhile I've found a way to make the ArcSoft DTS Decoder work inside of eac3to. However, I have to tell the decoder how many channels I want to have decoded. That's problematic because my DTS-HD channel detection isn't really reliable...

[yesgrey]

madshi,
My PC has 2 disks, and I usually work with eac3to from one disk to the other and vice-versa until all the processing needed is terminated. From my experience with eac3to it appears that some of the tasks use temp files.
I believe this should be a very simple thing to change, so is it possible for you to create some kind of setting to be used by people like me who uses two disks? It would be great for processing speed and also for disk protection.

For example:
Demuxing 1 video and 2 tracks for the raw streams took 9m16s.
Demuxing 1 video track to mkv and 2 audio tracks for raw streams took 11m18s.

It appears the video to mkv has some kind of temp file, I only don't know if it's located in the source disk or the target disk.

Thanks for your great work!

[rickardk]

Originally Posted by madshi

Yes.


Don't know if you can directly decode demuxed DTS files in GraphEdit with ArcSoft. Maybe it's possible somehow, not sure.

In the meanwhile I've found a way to make the ArcSoft DTS Decoder work inside of eac3to. However, I have to tell the decoder how many channels I want to have decoded. That's problematic because my DTS-HD channel detection isn't really reliable...

I re-ripped one movie with DTS-HD MA 7.1 audio. But still I can't connect to Arcsoft Audio Decoder HD. Do I have use more filters in the graph?

[nautilus7]

Originally Posted by madshi

In the meanwhile I've found a way to make the ArcSoft DTS Decoder work inside of eac3to. However, I have to tell the decoder how many channels I want to have decoded. That's problematic because my DTS-HD channel detection isn't really reliable...

Wow!! That's good news, after the bomb you dropped earlier! I am certain that we will get over any problems we come across.

I am quite sure I have dts-ma 4ch track (don't know if it's 4.0 or 3.1) but I can't find it right now. I guess that counts as "funny" format.

Originally Posted by rickardk

I re-ripped one movie with DTS-HD MA 7.1 audio. But still I can't connect to Arcsoft Audio Decoder HD. Do I have use more filters in the graph?

See 4-5 pages back, where xkodi posts his findings + graph.

[madshi]

Originally Posted by yesgrey3

For example:
Demuxing 1 video and 2 tracks for the raw streams took 9m16s.
Demuxing 1 video track to mkv and 2 audio tracks for raw streams took 11m18s.

It appears the video to mkv has some kind of temp file, I only don't know if it's located in the source disk or the target disk.

There is no temp file involved when doing MKV muxing. It's just that the MKV muxer costs a bit of performance.

Originally Posted by nautilus7

I am quite sure I have dts-ma 4ch track (don't know if it's 4.0 or 3.1) but I can't find it right now. I guess that counts as "funny" format.

Yes, it does. If you find, can you please upload a small sample (5MB is enough)? Thanks...

[ideas_man69]

just a quick question guys.. not sure if it has been asked - i've skimmed through quite a few pages of this thread already.

when converting tracks to DTS - is the general consensus that:

TrueHD to DTS > E-AC3 to DTS?

or

TrueHD to AC3 > E-AC3 to AC3?

TrueHD is decoded using ffmpeg, eac3 decoded by nero. Are nero's eac3 decoders better quality? Which mode retains the most information?

[EPiPH0NE]

Originally Posted by ideas_man69

just a quick question guys.. not sure if it has been asked - i've skimmed through quite a few pages of this thread already.

when converting tracks to DTS - is the general consensus that:

TrueHD to DTS > E-AC3 to DTS?

or

TrueHD to AC3 > E-AC3 to AC3?

TrueHD is decoded using ffmpeg, eac3 decoded by nero. Are nero's eac3 decoders better quality? Which mode retains the most information?

What I use for HD-DVD:

1. TrueHD -> PCM
2. TrueHD -> DTS
3. 640K/> EAC3 -> AC3 (core for BluRay)
4. 768k/< EAC3 -> DTS


BluRay:

1. TrueHD -> PCM
2. TrueHD -> DTS
3. TrueHD -> AC3 (core)
4. EAC3 (see HD-DVD)

IMHO, best quality is these orders. Depends on your hardware.

[G_M_C]

Originally Posted by madshi

eac3to v2.48 released

http://madshi.net/eac3to.zip

v2.48
* 96kHz LPCM tracks in (M2)TS and EVO/VOB containers didn't work correctly
* "Applying (E-)AC3 delay" now only shows if the bitstream is actually modified
* fixed crash in MP2 reader when checking some PCM tracks
* added support for MLP formats 13 - 16
* improved/corrected MLP channel descriptions
* MLP parser sets correct channel mask
* added proper channel remaps for libav MLP decoding of "funny" channel formats
* added proper channel remaps for Nero MLP decoding of "funny" channel formats
* added proper channel remaps for Nero AC3 decoding of "funny" channel formats
* when doubling 7th channel the channel mask is set correctly now
* channel mask is corrected if a decoder doesn't output all channels
* channel mask is corrected if a channel downmixing is performed

I've checked out the ArcSoft Audio Decoder. It seems to work well enough - but only when using it in GraphEdit and when manually adjusting the parameters to the correct values. When accessing the decoder from within eac3to the ArcSoft Audio Decoder always outputs stereo, only. I've not found a (documented/official) way how eac3to could convince the decoder to output more than 2 channels, sadly...

Hi Madshi,
I see that you havent made a reply to my post. Do you haveyou plans/time/possibillties to fullfill my request (converting to 24fps), or shall i simply keep using AviSynth + SoundOut () ?

[madshi]

Originally Posted by G_M_C

I see that you havent made a reply to my post.

Yes, I have. Look again...

[G_M_C]

Originally Posted by madshi

Yes, I have. Look again...

Oops, sorry i overlooked that. My bad ...



Regarding the 24 fps though; More and more displays are beinig "optimized" for 24 fps. Many newer display-manuafacturers even market those functions ("Real cinema 24fps" kind of features). My Panasonic 42pz85 is made for/accepts 1080p24 (and other 24 fps), and so are all displays from this years (2008) generation.

And not only displays are made to work on 24 fps, so are the (newer) players. My dpm dp30 has a specific setting to let it output at 24 fps; But also the PS3 has had an update to output @ 24 fps, and many other players recently been made fit to work on 24fps.

I do not know on what fps movies are encoded and/or stored on the physical disk, if it is 23,976 or not; But even PC-apps & rippers prodce 24 fps these days (and even the "less legal" full BD-rips / torrents often seem to be 24 fps). But now i'm rambling on ... but i think 24 fps will be seen more often in the future.

[tebasuna51]

Originally Posted by madshi

Do you happen to also have samples for all the funny DTS channel formats?

I make the samples for ac3/mlp with the encoders but with the Surcode DTS v1.0.23 only can make 5.0/5.1 samples and I can't check the eac3to detection and libav channelmap behaviour.

With less channels Surcode put empty channels. For instance:

eac3to v2.48
command line: "D:\Test\AudioN\eac3to\eac3to.exe" "E:\Test\wavs\2w200.wav" "E:\Test\wavs\2w200.dts"
------------------------------------------------------------------------------
WAV, 2.0 channels, 0:00:20, 16 bits, 48khz
Reading WAV...
Writing WAVs...
Creating file "E:\Test\wavs\2w200.L.wav"...
Creating file "E:\Test\wavs\2w200.R.wav"...
Found Surcode DTS Encoder version 1.0.23.0.
Surcode encoding successfully started. Please wait...
Closing Surcode...
eac3to processing took 1 second.
Surcode encoding took 6 seconds.
Done.


eac3to v2.48
command line: "D:\Test\AudioN\eac3to\eac3to.exe" "E:\Test\wavs\2w200.dts"
------------------------------------------------------------------------------
DTS, 5.0 channels, 0:00:20, 24 bits, 1536kbit/s, 48khz

With FC,BL,BR empty

Originally Posted by rickardk

Is the lfe channel always mixed at -10dB for all formats (AC3, E-AC3, TrueHD, DTS, DTS-HD (MA), LPCM)?

Does any of the default decoders apply the +10dB for the lfe channel? (I hope not)

I never see this fixed attenuation/amplification in encoders/decoders.

AFAIK, the encoders can use only (forced or optional) a lowpass filter for LFE channel. Other question is the recommended relative volume level between channels, but is a free choice for the soundtrack author.