eac3to - audio conversion tool

[nevcairiel]

Any ideas about this?

Blu-ray and DVD use different types of LPCM encoding, so with some confidence i can say that it won't be compatible with both at the same time.

[Richard1485]

Blu-ray and DVD use different types of LPCM encoding, so with some confidence i can say that it won't be compatible with both at the same time.

Thanks for the info. In that case, I need to know what commands to input for Blu-ray and for DVD.

[tebasuna51]

Thanks for the info. In that case, I need to know what commands to input for Blu-ray and for DVD.
That depend of the soft you want use to mux video and audio.
For instance tsMuxeR accept .wav (<4GB) and .w64 (>4GB) to obtain .m2ts files for BD

[Richard1485]

That depend of the soft you want use to mux video and audio.

Really? I understood that muxers don't actually change files: they just, well, mux them. I don't see how muxing .w64 can be BD compliant, unless tsmuxer changes the file in some way.

[mindbomb]

so, im not completely clear on the use of flac for things above 6 channels.
WAVEFORMATEXTENSIBLE_CHANNEL_MASK is supposed to be used? and eac3to does this by default?
And decoders should be able to use that for to figure out the channels?

If my ultimate goal is to use this in a matroska file, does the channel mask remain through the muxing process?

[tebasuna51]

Really? I understood that muxers don't actually change files: they just, well, mux them. I don't see how muxing .w64 can be BD compliant, unless tsmuxer changes the file in some way.

Of course, a muxer can modify the data order (unique diference between lpcm and wav/w64 files)

so, im not completely clear on the use of flac for things above 6 channels.
WAVEFORMATEXTENSIBLE_CHANNEL_MASK is supposed to be used? and eac3to does this by default?
And decoders should be able to use that for to figure out the channels?

If my ultimate goal is to use this in a matroska file, does the channel mask remain through the muxing process?

AFAIK, flac can't store channel mask.
There are a unique channel mask supported for each num_channels type, and undefined for 7 and 8 channels.

[me7]

eac3to doesn't seem to like 5.0 content. I have a 5.0 DTS Master Audio tracks that I want to encode to 5.0 AC3, but eac3to crashes with an error:

DTS Master Audio, 5.0 channels, 24 bits, 48kHz
(core: DTS, 5.0 channels, 24 bits, 1509kbps, 48kHz)
Decoding with ArcSoft DTS Decoder...
The AC3 encoder received a non-supported data format (pcm, 5, 24, -).

Apparently the included AC3 encoder doesn't like 5 channel content.
If I use eac3to to create a 5.0 wav file and encode it with a standalone AC3 encoder, do I need to worry about channel order?

[rapscallion]

Audio question about my logic :

If I extract a 385 Mb AC-3, 5.1, 640 Kbps track to 6 wavs, the results are 975 Mb wavs @1152 Kbps each.

Enter the wavs into DTS-MA suite, encode them into DTS-High Res audio track format with a bitrate @ 3840 Kbps.
The result is still lossy ( I know they can't be converted to DTS-HD) but much higher bitrate and, I would think, higher quality than DTS @ 1509 Kbps and the original AC-3 track.

Is my reasoning correct or am I out in left field somewhere ?

[Asmodian]

... higher quality than DTS @ 1509 Kbps and the original AC-3 track.

You can never get higher quality than the original, a lossy reencode will always be lower quality than the source.

How much lower is dependent on the settings and in your case I would assume not much. Still it is better to just leave it as the AC3 if you can.

[Richard1485]

Of course, a muxer can modify the data order (unique diference between lpcm and wav/w64 files)

Ah, so the only difference between those types of files is data order, and as long as the muxer accepts the input the result should be compliant. Thanks. That helps.

[rapscallion]

You can never get higher quality than the original, a lossy reencode will always be lower quality than the source.

How much lower is dependent on the settings and in your case I would assume not much. Still it is better to just leave it as the AC3 if you can.

However, the AC-3 isn't the original, the wavs are. Thus, a re encode of the extracted wavs should result in better quality at the higher bitrate, no?

Edit: This stuff really does make my head hurt.
BTW, I did the encode and the resulting DTS-HR file is ~ 3.2 gb

[tebasuna51]

...
If I use eac3to to create a 5.0 wav file and encode it with a standalone AC3 encoder, do I need to worry about channel order?

No problem.
You can use also eac3to with 'pipe' if channels aren't 2.0 or 5.1:
eac3to input stdout.wav | Aften -b 640 -readtoeof 1 output.ac3

However, the AC-3 isn't the original, the wavs are.

The decoded ac3 wav's aren't the originals, have the same quality than the ac3.

[Asmodian]

If I extract a 385 Mb AC-3, 5.1, 640 Kbps track to 6 wavs

However, the AC-3 isn't the original, the wavs are.


If you started with it as your source of the audio the AC3 is your "original". You cannot recover the real original audio; some information was lost when it was compressed to AC3.

As tebasuna51 said the wav files contain the same information as the AC3, just no longer in a compressed form.

[rapscallion]

If you started with it as your source of the audio the AC3 is your "original". You cannot recover the real original audio; some information was lost when it was compressed to AC3.

As tebasuna51 said the wav files contain the same information as the AC3, just no longer in a compressed form.
Yes, what I should have said is that the wav files are the originals minus the "unnecessary" data that the compression expunges.

So, am I to understand that a 5.1 AC-3 640 Kbps audio is the same quality as it's 1150 Kbps extracted wavs? If that's the case, then shouldn't a 440 Kbps be the same quality as a 640 Kbps if compressed form the same source ?
I just can't get my head around this, so sorry for dwelling on it.

[Asmodian]

It is lossy compression, an AC3 is not like a zip file where after decompression you get exactly the same data back. Lossy compression drops information it thinks you will not miss (much); the 440 Kbps file dropped more information than the 640 Kbps file. Both dropped some information compared to the original.

It isn't really "unnescessary" data it is just less noticable.

I should also mention that re-encoding your extracted wavs to a 640 Kbps AC3 would lose information again; your new AC3 would be lower quality compaired to the original AC3.

[rapscallion]

Ok, thanks Asmodian. No how much I googled this, I couldn't find a concise answer.
Especially re the greater the compression, the more information dropped. So the answer to my previous question "5.1 AC-3 640 Kbps audio is the same quality as it's 1150 Kbps extracted wavs? " is YES. Very interesting.

[Asmodian]

So the answer to my previous question "5.1 AC-3 640 Kbps audio is the same quality as it's 1150 Kbps extracted wavs? " is YES. Very interesting.

You can trust tebasuna51. ;)
We both answered this in the previous two posts.

When you play an AC3 (or any audio file) it is converted to raw digital audio data and sent to your sound card for digital to analogue conversion. "Extracting" to a wav file is like playing the AC3 but saving the digital data to the hard drive instead of sending it to the sound card.

[rapscallion]

I know you did, I was just confirming that I now get it.

[Asmodian]

Sorry, I thought you were annoyed I didn't answer your obvious question. I hate ignored questions but left that one out as it was already addressed.

Glad it makes sense now. :)

[odin24]

Are there any issues converting a 7.1 DTS-HD MA to DD5.1, in regards to the back channels mixing into the surround channels? I use Arcsoft to decode the audio, output to .w64, encode to DD5.1 (libav).

[Sparktank]

Are there any issues converting a 7.1 DTS-HD MA to DD5.1, in regards to the back channels mixing into the surround channels? I use Arcsoft to decode the audio, output to .w64, encode to DD5.1 (libav).

According to tebasuna51, there shouldn't be any problems downmixing 7.1 DTS-HD MA to 5.1 (http://forum.doom9.org/showthread.php?p=1529424#post1529424).
As long as you're using Arcsoft 1.1.0.0 (25/04/2008), it can downmix 7.1 with either proper channel setup or strange setup.
Post 1 (http://forum.doom9.org/showthread.php?p=1519949#post1519949); Post 2 (http://forum.doom9.org/showthread.php?p=1519992#post1519992); Arcsoft version info (http://forum.doom9.org/showthread.php?p=1266679#post1266679)

Not quite sure if I missed anything... :scared:
There is this post (http://forum.doom9.org/showthread.php?p=1520511#post1520511) of some interest.

[taiyoyuden]

For best DTS decoding you need:
(1) ArcSoft DTS Decoder - version 1.1.0.0 or newer

Is the ArcSoft DTS Decoder 1.1.0.0+ the recommended/default DTS-HD decoder now? I'm confused because here (http://en.wikibooks.org/wiki/Eac3to/In_Depth_Technical_Explanation#Evaluation_of_available_decoders), the default says Sonic and no mention of the Arcsoft. Thus, I've been using the Sonic 4.3.0.169 decoder for numerous DTS-HD 5.1 tracks.

Is there any difference in sound quality, channel mapping, etc.? Should I have them redone with the ArcSoft? Which version of the ArcSoft DTS Decoder is the best?

[tebasuna51]

Please read the first post in this thread:

"The Sonic DTS decoder is very good for DTS, DTS-ES, DTS-96/24, DTS-HD Master Audio and DTS-HD High Resolution tracks. The only problem is that it decodes DTS-HD 7.1 tracks only as 5.1.
...
The ArcSoft DTS decoder seems to be perfect for DTS and DTS-HD decoding. It supports every format and channel configuration that exists including 6.1 and 7.1."

I recommend ArcSoft DTS Decoder 1.1.0.0

[Richard1485]

Tebasuna, taiyoyuden also asked if there is any difference in sound quality between ArcSoft and Sonic when DTS-HD 5.1 tracks are being decoded. I am interested in knowing the answer to this too. The only limitation for Sonic mentioned on the first page is that 7.1 tracks are decoded as 5.1.

[tebasuna51]

Decoding lossless (DTS-MA) don't exist differences.
Decoding lossy DTS all decoders can have very little differences but is dificult to say what is better.
Even free decoders are good for standard DTS.

[Richard1485]

Thank you.

[Marin85]

I am wondering about the precise meaning of the following type of log messages from eac3to: "A remaining delay of +1ms could not be fixed." Does it mean that the audio track still has a residual delay of +1ms and hence I hypothetically need to somehow apply -1ms delay to fix it completely if I want to? Or does it mean that eac3to managed to apply only (X-1)ms delay, where -X is the full audio delay determined by eac3to, and I still need to apply +1 ms (hypothetically) delay to the audio track to fix it completely?

This is just a theoretical question. I am well aware that in general I should not be worrying about 1ms audio delays, nor that I would be able to fix such a small value via eac3to (say, for an AC3 audio track). I would be thankful for a straight and clear answer. I have searched the web and there appear to be some controversial opinions on this matter.

[pandv2]

AC3 is a packed format. What it means is the samples are grouped and compressed. The packet lenght and duration is fixed and depends on the samplerate (44.1 KHz, 48 KHz...). If you want to do a lossless operation with a AC3 stream you need to do at packet level. Supose (I don't remember the real number) a packet is 16 ms, you want to delay the stream 20 ms, eac3to only can add a complete ac3 silence packet (16 ms) or two (32 ms). In the first case the error is -4ms and in the second +12ms.

[Marin85]

AC3 is a packed format. What it means is the samples are grouped and compressed. The packet lenght and duration is fixed and depends on the samplerate (44.1 KHz, 48 KHz...). If you want to do a lossless operation with a AC3 stream you need to do at packet level. Supose (I don't remember the real number) a packet is 16 ms, you want to delay the stream 20 ms, eac3to only can add a complete ac3 silence packet (16 ms) or two (32 ms). In the first case the error is -4ms and in the second +12ms.
Thank you for your reply, but my question is not specifically about delays in AC3 tracks, but about how to interpret a particular type of messages in the eac3to log.

[tebasuna51]

Or does it mean that eac3to managed to apply only (X-1)ms delay, where -X is the full audio delay determined by eac3to, and I still need to apply +1 ms (hypothetically) delay to the audio track to fix it completely
That is correct.

[Boulder]

Regarding audio delay: if DGIndex tells me that the audio track has +144ms delay, do I need to use +144ms or -144ms in eac3to's command line? It's a silly question but there is no obvious answer ;)

[Marin85]

Regarding audio delay: if DGIndex tells me that the audio track has +144ms delay, do I need to use +144ms or -144ms in eac3to's command line? It's a silly question but there is no obvious answer ;)
I feel like I am being mocked here :) What tebasuna51 (btw, thank you for the clear answer!) explained is - as I suspected - sort of the opposite of what eac3to is actually suggesting by the above message IMHO. It is confusing because eac3to message refers to the delay it tries to apply to the audio track, and not to the (resulting) delay the audio track already has.

[Boulder]

I feel like I am being mocked here :)Not at all :) It's just very confusing at times. I think I've always used the opposite but since the delays are usually rather small, it's not easy to determine which way is the correct one.

[tebasuna51]

... if DGIndex tells me that the audio track has +144ms delay, do I need to use +144ms...
Yes, you must always respect the sign.
The info (also MediaInfo) always show the delay than you need apply.

[tormento]

There is a stream that eac3to really doesn't like:

9: DTS Express, English, 1.0 channels, 24 bits, 96kbps, 48kHz

I can demux it but nothing more. Tried with TotalMedia libraries, LibAV ones but whenever I am going to export to wavs, it tells me some kind of errors. Any idea about how to convert it?

[Brazil2]

There is a stream that eac3to really doesn't like:

9: DTS Express, English, 1.0 channels, 24 bits, 96kbps, 48kHz

I can demux it but nothing more. Tried with TotalMedia libraries, LibAV ones but whenever I am going to export to wavs, it tells me some kind of errors. Any idea about how to convert it?
A sample would be nice for testing purposes :)

[tormento]

A sample would be nice for testing purposes :)
If only could be possible to demux it somehow..

[Brazil2]

If only could be possible to demux it somehow..
Well, you said you did:
I can demux it but nothing more.


Anyway, post a muxed sample so we can check it ?

[Bryce2]

Hi to all!
Trying to convert a DTS track to AC3 using "eac3to" leads me to an unusual problem.
DTS (ES):2.77GiB, 4h 23mn = AC3: 1.34GiB, 5h 0mn.
I want to mention that this is only the 2nd time I'm facing this problem after more than 200 conversions. The first time, I found a solution in this forum with adding a parameter to eac3to, but now I'm searching hours & hours with no luck.

[Sparktank]

Hi to all!
Trying to convert a DTS track to AC3 using "eac3to" leads me to an unusual problem.
DTS (ES):2.77GiB, 4h 23mn = AC3: 1.34GiB, 5h 0mn.
I want to mention that this is only the 2nd time I'm facing this problem after more than 200 conversions. The first time, I found a solution in this forum with adding a parameter to eac3to, but now I'm searching hours & hours with no luck.

Can you post the full log?

[Bryce2]

Hi to all!
Trying to convert a DTS track to AC3 using "eac3to" leads me to an unusual problem.
DTS (ES):2.77GiB, 4h 23mn = AC3: 1.34GiB, 5h 0mn.
I want to mention that this is only the 2nd time I'm facing this problem after more than 200 conversions. The first time, I found a solution in this forum with adding a parameter to eac3to, but now I'm searching hours & hours with no luck.

Solved:
Using as usual the default command "eac3to.exe input.dts output.ac3" was the reason of this weird effect (for this & only dts stream).
After trial & error and changing the command to "eac3to.exe input.dts output.ac3 -libav" ..did the thing.
Anyway ..:thanks: "Sparktank" for your interest to give a help.

[DJ-1]

Can I convert maintain DTS-MA track using eac32 ?
Sent from my GT-I9100 using Tapatalk 2

[DJ-1]

Oops, double post

[sneaker_ger]

eac3to can do that with the help of external libraries like Arcsoft. Read the start post.

[DJ-1]

eac3to can do that with the help of external libraries like Arcsoft. Read the start post.
OK, will read it, I was trying out Multi avchd Wichita uses eac32.... tried converting dts-ma to LPCM, (as my media box does play dts, but not .dts-ma..... ) so I was trying g to passtgrough the pre-decoded LPCM stream.. got lots of hissing...nothing more.
I have an Onkyo Tx nr609, (does support dts-ma )
Sent from my GT-I9100 using Tapatalk 2

[frumble]

Hello,
I have a problem with eac3to that drives me crazy. I want to decode the DTS-HD MA streams of my release of x and encode them to FLAC so that my Linux audio players can play the full sound, not just the DTS core.
I have downloaded eac3to and the ArcSoft DTS Decoder 1.1.0.0 plus the HdBrStreamExtractor GUI.
The output WAVE files (or FLACs) have all 7 channels and the sound is nice but they appear to have the wrong channel mapping: Voices are very low and often come from only one side (I have only stereo speakers) and once in a while come with strange echo. But when I import them into Audacity and play them with the editor the output is correct. This however has no effect on the Audacity export: 6.1 channel export has the same mapping problems as the original file. My players are not the problem, they can play every DTS, AC3 and TrueHD stream correctly.
I read something about "strange setup" but I can't understand if this might be the problem. I have absolutely no knowledge about audio mastering. The issue affects both English and German streams. The Audacity export window offers the option to remap the channels but I have no clue what could be the right choice. So I really hope you can help me: I uploaded a piece of 6.1 export from Audacity in Ogg Vorbis and it would be really nice if someone could tell me the right mapping. Thank you very much in advance!
http://www.2shared.com/audio/boFrWT5B/Export-example-eac3to-wrong-ch.html - I hope not to violate against forum rules with this.

Off this topic I want to say something: It may sound ridiculous but I miss a "tutorial". I lost hours in trying to understand what these freeware tools do and how to set them up with decoder and codecs. The learning curve is very steep but not because I am a idiot but 'cause most of your tools lack a proper documentation. I am a Linux user and I am used to find answers in reading docs.

It's really great that there are freeware tools like eac3to but I can't understand why the authors of such media helpers don't make the source code available. The last version of eac3to is from 2010? I read something about a GPU video encoder the author is writing since then and the promise to get back to work on eac3to when this encoder is final. But since 2010 no progress with eac3to. In this time the program could have been matured and bugs could have been fixed from others but they can't do it because they don't have the source code. It is his right to hold the source for himself but I truly can't understand the reason. It doesn't seams to be his plan to make commercial profit out of it. But no hard feelings, this are just my thoughts.

[NanoBot]

Hi frumble,

this workaround might help: http://forum.doom9.org/showthread.php?p=1524301#post1524301

[sneaker_ger]

Unfortunately 6.1 is not defined in FLAC. So players might decode it not as you expect, unless you do a downmix.

In addition to NanoBot's link:
http://forum.doom9.org/showthread.php?p=1538930#post1538930

[Sparktank]

Unfortunately 6.1 is not defined in FLAC.

This has actually been a bit of a constant problem for FLAC.
One user here goes on about it...
Dear (people), please stop using 6.1 FLAC (http://anonym.to/?http://mod16.org/hurfdurf/?p=184)

I remember reading about 6.1 FLAC in other places.

Personally, I'd stick to the CORE or even a conversion to AC3@640, as 640 is perceived to have transparent quality to the master file.

With some BD rips I do, I'm quite happy using only the core for maximum compatibility with my BD player that can playback MKV (with original h264/VC1/MPEG).

It's not a significant compromise between lossless and core, I'd stick to CORE audio. The human ear wouldn't really be able to pick pu the difference too much.

[Midzuki]

Another possible approach: drop FLAC :p and "move house" to WavPack :)