eac3to - audio conversion tool

[nautilus7]

It should be the wav header in the begging of each file.

[tebasuna51]

Originally Posted by piratburner

OK I use Pcm2Tsmu for my pcm track, but adding the file to TsMexer, then is telling me thats is a 5.1 track. And playing the m2ts file is giving me funny nois. Is it a bug in eac3to when converting to PCM

Eac3to don't have a bug please go to Pcm2Tsmu thread.

[Snowknight26]

8: AC3, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
"Commentary By Director Jonathan Demme And Screenplay Co-Writer Daniel Pyne"
[...]
[a08] Reducing depth from 64 to 24 bits...

When converting to wav.

[Thunderbolt8]

a technical question: when comparing 2 source.track -> .wav files of the same channel from different source types (e.g. center channel of 5.1 ac3 and flac or left and right channel of 2.0 ac3 with left and right channel of 5.1 flac) and looking for their different delays is it technically correct then to look where the first tiny bit of sound occurs in the channel graph? what I mean is that a flac track of course has a higher quality as an ac3 track and this difference must be located somewhere. so could it be possible that the flac track has numerical more digits in the graphs at the beginning of the track compared to the ac3 track so that this kind of delay meassurement is wrong, or is this difference of sound quality only to be found in the amplitude of the graph, that it only spread in the vertical, but not the horizontal direction of the graph?

[svgame]

thank you Madshi
just there has a problem.i make 3 step
1) use eac3to demux TrueHD track directly from BD/HD
2) type command "eac3to 00001.thd audio.thd" "eac3to 00001.thd audio.ac3",demuxing to ac3 core and lossless part
3) rename audio.thd to audio.mlp,and import to Scenarist.at 100%,its hung!!!about 2 minute，get a error message "External component has thrown an exception"
where is problem?Madshi

[ACrowley]

Originally Posted by tebasuna51

I have a method to extract the BC channel but is not easy, work fine with test samples (channels separated in time) but I can test real samples. Maybe you can make the test:
- Decode the 6.1 matrix to 6 monowavs (5.1)
- Merge (WaveWizard or Sox) the Back (or Side) channels to stereo.
- Decode this stereo file like DolbyPrologic with Foobar2000 and the Free Surround plugin decoder (Center=1, Dimension=-0.5, A=0, B=0).
- The FL,FR,FC from the resultant 5.1 output are your BL,BR,BC channels.

With the test file:
FL = 0.9448 x BL + 0.0004 x BC
FR = 0.9448 x BR + 0.0004 x BC
FC = 0.0235 x BL + 0.0248 x BR + BC
LF < 0.0000
SL < 0.0000
SR < 0.0000



I don't know what is my problem, I'm wait to madshi can see anything at bug report.

so the outout is identical to the source BC Channel ?
Will i twork with DD EX tracks?

By the Way, i compared all Surround/Back Channles from dts es discrete with the test matrix reencode and arcsoft -7.
Theyre all looking fine and identical compared to the original dts es discrete....

However, Madshi will make it ,as always

[tebasuna51]

Originally Posted by ACrowley

so the outout is identical to the source BC Channel ?

The analog extraction isn't perfect, like you see there are BL and BR parts (2%) added to the BC channel.

Will i twork with DD EX tracks?

Seems works with dolby-waterfal_51EX sample but I don't have channel test EX to quantify the separation.

By the Way, i compared all Surround/Back Channles from dts es discrete with the test matrix reencode and arcsoft -7.
Theyre all looking fine and identical compared to the original dts es discrete....

Then if work for you don't need nothing more.

[ACrowley]

Originally Posted by tebasuna51

Then if work for you don't need nothing more.

yeah..it should be tested more..im not sure
Strange, i cant reproduce my Test. eac3to/arcsoft -7 crashes on the dts es matrix reencode. But it works with other matrix dts tracks!

Madshi should take a closer look, he is the master

[rica]

Hi guys.

How can i shorten this:

eac3to  E:\VIDEO_TS\VTS_01.vob+E:\VIDEO_TS\VTS_01_2.vob 2: C:\KB\video.m2v 6: C:\KB\dtsaudio.dts 5:

This:

eac3to E:\VIDEO_TS\[VTS_01_1+VTS_01_2].vob

or this:

 eac3to E:\VIDEO_TS\[VTS_01_1.vob+VTS_01_2.vob]

doesn't work?


_ _ _ _

[madshi]

Originally Posted by BlackJack1

I need to remux no. 4: DTS-ES 6.1 track 1536k and 3: DD 5.1EX 640k track.
Here is the log:

but... PDVD reports that I've got DTS-ES 5.1 1536k track and not 6.1...
Previously during playing BD movie using the same PDVD - reported: DTS-ES 6.1 1536k.
Please explain what I did wrong.

Don't trust in what PDVD says. Run "eac3to demuxedtrack.dts". What does eac3to say?

Originally Posted by itsancho

btw, madshi, is there any chance eac3to to report if AC3 track is DD-EX?

Will add that to my to do list.

Originally Posted by tebasuna51

Not all Ok here.
If we name source channels like Rear: RL, RR, RC and

RL' = +90º phase shift RL
RR' = -90º phase shift RR

seems the phase shift is make at encoder phase and is the same for 'matrix' and 'discrete' DTS-ES in all the tests.

Ouch. Don't you consider that "bad"? I mean just imagine we want to speed correct (e.g. apply PAL speedup) to a DTS track. Doesn't that mean that the phase changes with every new reencode? So if I do "eac3to source.dts dest.dts -speedup" the phase of the surround channels will be different?

Originally Posted by tebasuna51

1) 3/3.1 -> 5.1 (like previous test)

SL = RL' + 0.71 x RC
SR = RR' + 0.71 x RC
Ok

Ok, that's fine, that's how eac3to is also doing 6.1 -> 5.1 downmixing.

Originally Posted by tebasuna51

3) 3/3.1 -> 7.1

SL = 0.40 x RL' + 0.60 x RC
SR = 0.40 x RR' + 0.60 x RC
BL = 0.75 x RL'
BR = 0.75 x RR'
Seems wrong: RC only in Side and most Rear part in Back, I think must be the other way, Side <-> Back

Agreed.

Originally Posted by tebasuna51

2) 3/3.1 -> 7.1

SL = 0.71 x RL' + 0.50 x RC
SR = 0.71 x RR' + 0.50 x RC
BL = 0.71 x RL' + 0.50 x RC
BR = 0.71 x RR' + 0.50 x RC
For me seems ArcSoft can't recover the RC mixed in RL-RR because RC must be present only in Back channels.

Agreed.

Originally Posted by tebasuna51

3) 3/3.1 -> 6.1, eac3to crash with this bugreport.txt

The report was made with 2.70 but is the same with 2.72, I send the 2.70 bug because was made after a reset with the minimum of soft running.

It seems that the ArcSoft decoder crashes internally somewhere. That is probably a bug in the ArcSoft decoder. You have version 1.1.0.0.

@ACrowley, are you using a different ArcSoft decoder version? FWIW, I'm getting the same crash as tebasuna51...

Originally Posted by himan2001

Still the same "issue" with DTS and the -core option.

I personly dislike 7.1 downmixing from DTS-HD 7.1 files, when
a "clean" 5.1 core is present.

But is not possible to reencode directly in 1 step to ac3 from the -core. eac3to always sees the 7.1 MA and try to downmix from the extention. At the moment, first the -core must be exported, than you can ac3-encode from the 5.1 core.

There's a bug in eac3to. If you use the "-core" option, eac3to nevertheless thinks that it will get 7.1 from the decoder. Will have to fix that.

However, I'd strongly recommend that you reconsider your transcoding method. The "clean" 5.1 core will result in worse audio quality than the eac3to produced 7.1 downmix because the 7.1 source has a higher quality in every single channel compared to the core 5.1 track. Even if you're afraid that eac3to's 7.1 -> 5.1 downconversion is worse compared to what the studio did, still the front channels and LFE will have a higher quality if you use the full DTS-HD track. So IMHO using only the core is a bad idea.

Originally Posted by Thunderbolt8

Got a problem with a H.264 TV cap:
(actually wanted to slow it down, both video and audio, to 23.976 fps, but this doesnt work as well ofc)

eac3to v2.72
command line: eac3to G:\bla.ts G:\bla.mkv
------------------------------------------------------------------------------
TS, 1 video track, 1 audio track, 0:00:05
1: h264/AVC, 1080i50 (16:9)
2: AC3, English, 5.1 channels, 384kbps, 48khz, dialnorm: -27dB, 57ms
[v01] Extracting video track number 1...
[a02] Extracting audio track number 2...
[v01] Muxing video to Matroska...
[a02] Removing AC3 dialog normalization...
[a02] Applying (E-)AC3 delay...
[a02] This doesn't seem to be a valid PES packet.
Aborted at file position 1277952.

heres a 10mb sample: http://www.sendspace.com/file/yveygw

Looks like a broken/corrupted file.

Originally Posted by Thunderbolt8

does anyone know how to create a single channel wav file (only the center channel) of a multi channel flac/ac3/lpcm/dtshd/truehd source track?
when I do "eac3to source.track outcome.wav" then I only get one huge file, which includes all the source channels and not only the center one I want to have.

This will decode the first 50MB of the source file and store only the center channel into a WAV file:

"eac3to sourcefile test.wav -mono -50mb"

Originally Posted by Thunderbolt8

a technical question: when comparing 2 source.track -> .wav files of the same channel from different source types (e.g. center channel of 5.1 ac3 and flac or left and right channel of 2.0 ac3 with left and right channel of 5.1 flac) and looking for their different delays is it technically correct then to look where the first tiny bit of sound occurs in the channel graph?

I'd suggest using Audacity or a similar wave editor and compare the form of the graphs. It's very easy to see how much delay an audio track needs this way.

Originally Posted by MichaelAnders

Using v2.72, it seems as if the channel order is corrupt?

M2TS, 1 video track, 2 audio tracks, 1:21:00
1: VC-1, 1080p24 /1.001 (16:9)
2: DTS Master Audio, German, 5.1 channels, 24 bits, 48khz
3: DTS Master Audio, English, 7.1 channels, 16 bits, 48khz
[a03] Extracting audio track number 3...
[a03] Decoding with ArcSoft DTS Decoder...
[a03] Encoding FLAC with libFlac...
[a03] Creating file "e.flac"...
[a03] The original audio track has a constant bit depth of 16 bits.

If I convert the German audio to FLAC (same as before, this time only the 2nd stream), everything is fine and the file sounds perfect in MPCHC and also in foobar2000:
[a02] Original audio track: max 24 bits, average 16 bits, most common 16 bits.

The English FLAC however, is totally wrong. It seems as if FL and FR are now SL and SR, the subwoofer is also on some other channel...

If I play the English stream in the M2TS file, the "properties" in MPCHC shows "Audio: DTS 48000Hz 6ch [Audio]".

I don't have any older version of eac3to to check this, but something is wrong... Maybe eac3to finds 8 channels, but it's actually 6?

Please try "eac3to English.flac Englisch.wav -50mb". That will decode the first 50MB of the FLAC file to a WAV file. Then load this WAV file in e.g. Audacity and check whether the channel order is correct or not.

Or if you want me to check it, please upload a small sample of the original 7.1 English DTS track, please?

Originally Posted by Snowknight26

8: AC3, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
"Commentary By Director Jonathan Demme And Screenplay Co-Writer Daniel Pyne"
[...]
[a08] Reducing depth from 64 to 24 bits...

When converting to wav.

Are you trying to test whether I can read your mind?

Originally Posted by svgame

thank you Madshi
just there has a problem.i make 3 step

Please stop asking the same question over and over again. I've already answered it a while ago.

[madshi]

Originally Posted by rica

How can i shorten this

By calling eac3to from inside "E:\VIDEO_TS". Then you don't need to use any paths for the source files.

Originally Posted by Octo-puss

I got a question. Is there any room for performance improvement? It looks to me like the data extracting is pretty slow (I don't understand how it works though, so it might be normal of course). The other day I took one of my movies and dug the AC3 audio out of it - it wasn't even converting from other format - and it was pretty slow...

When extracing one AC3 track from a (M2)TS file I'm getting about 13-15 MB. That's ok, I think?

[rica]

Originally Posted by madshi

By calling eac3to from inside "E:\VIDEO_TS". Then you don't need to use any paths for the source files

Thanks madshi but this time i get this:

C:\>eac3to E:\VIDEO_TS VTS_01_1.vob+VTS_01_2.vob
HD DVD / Blu-Ray disc structure not found.

[Snowknight26]

Originally Posted by madshi

Are you trying to test whether I can read your mind?

No, but I didn't think a sample was necessary. I was able to convert it to wav using a previous version (can't remember which), but it doesn't work ever since you added the deeper bitdepth analysis, leading me to believe its a small bug introduced in your code only recently.

Originally Posted by madshi

When extracing one AC3 track from a (M2)TS file I'm getting about 13-15 MB. That's ok, I think?

I have a server capable of doing 750MB/s average read with a C2Q, but it always takes eac3to about 10 minutes to demux a audio stream from your average length movie. Lets hope there is room for improvement.

[madshi]

Originally Posted by rica

Thanks madshi but this time i get this:

C:\>eac3to E:\VIDEO_TS VTS_01_1.vob+VTS_01_2.vob
HD DVD / Blu-Ray disc structure not found.

I think you need to learn how to use command lines...

You should have "E:\VIDEO_TS\>eac3to VTS_01_1.vob+VTS_01_2.vob".

Originally Posted by Snowknight26

No, but I didn't think a sample was necessary.

I'm not asking for a sample. You posted a log (which doesn't contain any error messages) without any further comment or question. What am I supposed to do with the log? I don't even know what you want from me.

[Snowknight26]

Originally Posted by madshi

I'm not asking for a sample. You posted a log (which doesn't contain any error messages) without any further comment or question. What am I supposed to do with the log? I don't even know what you want from me.

[a08] Reducing depth from 64 to 24 bits...

Since when are AC3 tracks 64 bit?

[rica]

Originally Posted by madshi

I think you need to learn how to use command lines...

You should have "E:\VIDEO_TS\>eac3to VTS_01_1.vob+VTS_01_2.vob".

Agree

[madshi]

Originally Posted by Snowknight26

Since when are AC3 tracks 64 bit?

AC3 tracks decode to floating point. The latest eac3to version internally uses 64bit floating point now.

[Snowknight26]

Ah, my mistake then. Good to hear though.

[tebasuna51]

Originally Posted by madshi

Ouch. Don't you consider that "bad"? I mean just imagine we want to speed correct (e.g. apply PAL speedup) to a DTS track. Doesn't that mean that the phase changes with every new reencode? So if I do "eac3to source.dts dest.dts -speedup" the phase of the surround channels will be different?

Yes I think is "bad".
I don't know if DTS PRO ENCODER have a switch to apply or not the phase shift to surround channels (like Dolby Encoders have)

[MichaelAnders]

Originally Posted by madshi

Or if you want me to check it, please upload a small sample of the original 7.1 English DTS track, please?

I don't have the tool you talked about, so here is the DTS file. I demuxed it via tsmuxergui, just the first 90 seconds. Interestingly, tsmuxergui also says it is 6 channels...

http://www.sendspace.com/file/6a3cl7