eac3to - audio conversion tool

[LigH]

Regarding DVD Video, that was one of the reasons to extract "the movie PGC" instead of processing VOB sequences as authored: to avoid trailer audio issues like asynchronity.

Removing trailers from BD playlists may be a different topic...

[Thunderbolt8]

are DTS:X headphone (2.0) tracks which are currently detected as DTS track by eac3to actually lossy or lossless tracks, like other DTS-HD MA tracks? and are DTS:X tracks are not denoted as such by eac3to yet?

[Zenitram]

Originally Posted by Thunderbolt8

DTS:X headphone (2.0) trac

Could you provide a sample file of DTS:X headphone?

[ndjamena]

If anyone is interested here's the latest unofficial MediaInfo dll:

http://www.mediafire.com/download/5y.../MediaInfo.zip

This is what it gets you:

General
Unique ID                                : 219291506723631580012532402152276823948 (0xA4FA02080541F5119808D773964D0F8C)
Complete name                            : D:\Zilla.mkv
Format                                   : Matroska
Format version                           : Version 4 / Version 2
File size                                : 36.8 GiB
Duration                                 : 2h 18mn
Overall bit rate mode                    : Variable
Overall bit rate                         : 37.9 Mbps
Movie name                               : Zilla
Encoded date                             : UTC 2015-11-17 09:05:56
Writing application                      : mkvmerge v8.5.1 ('Crosses') 64bit
Writing library                          : libebml v1.3.3 + libmatroska v1.4.4

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : High@L4.1
Format settings, CABAC                   : Yes
Format settings, ReFrames                : 2 frames
Format settings, GOP                     : M=1, N=10
Codec ID                                 : V_MPEG4/ISO/AVC
Duration                                 : 2h 18mn
Bit rate mode                            : Variable
Bit rate                                 : 34.7 Mbps
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 23.976 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.698
Stream size                              : 33.6 GiB (92%)
Language                                 : English
Default                                  : No
Forced                                   : No

Audio
ID                                       : 2
Format                                   : FLAC
Format/Info                              : Free Lossless Audio Codec
Codec ID                                 : A_FLAC
Duration                                 : 2h 18mn
Bit rate mode                            : Variable
Bit rate                                 : 3 202 Kbps
Channel(s)                               : 6 channels
Channel positions                        : Front: L C R, Side: L R, LFE
Sampling rate                            : 48.0 KHz
Bit depth                                : 24 bits
Detected bit depth                       : 20 bits
Stream size                              : 3.10 GiB (8%)
Writing library                          : libFLAC 1.2.1 (UTC 2007-09-17)
Language                                 : English
Default                                  : Yes
Forced                                   : No

Text #1
ID                                       : 3
Format                                   : PGS
Muxing mode                              : zlib
Codec ID                                 : S_HDMV/PGS
Codec ID/Info                            : Picture based subtitle format used on BDs/HD-DVDs
Duration                                 : 2h 1mn
Bit rate                                 : 11.4 Kbps
Count of elements                        : 2084
Stream size                              : 9.90 MiB (0%)
Language                                 : English
Default                                  : No
Forced                                   : No

Text #2
ID                                       : 4
Format                                   : PGS
Muxing mode                              : zlib
Codec ID                                 : S_HDMV/PGS
Codec ID/Info                            : Picture based subtitle format used on BDs/HD-DVDs
Duration                                 : 1h 23mn
Bit rate                                 : 198 bps
Count of elements                        : 36
Stream size                              : 121 KiB (0%)
Language                                 : English
Default                                  : No
Forced                                   : No

Text #3
ID                                       : 5
Format                                   : PGS
Muxing mode                              : zlib
Codec ID                                 : S_HDMV/PGS
Codec ID/Info                            : Picture based subtitle format used on BDs/HD-DVDs
Duration                                 : 2h 9mn
Bit rate                                 : 7 053 bps
Count of elements                        : 2460
Stream size                              : 6.52 MiB (0%)
Language                                 : English
Default                                  : No
Forced                                   : No

Menu
00:00:00.000                             : en:Start
00:07:48.092                             : en:Collision Course
00:17:25.502                             : en:Worm Guy
00:25:28.026                             : en:"Gojira"
00:36:53.461                             : en:Inside a Footprint
00:41:47.880                             : en:Insurance Claim
00:48:19.479                             : en:Caught on Something
01:01:35.900                             : en:"I Got a Bite"
01:11:05.803                             : en:New Kid in Town
01:21:38.685                             : en:23Rd St. Station
01:30:26.087                             : en:Drawing Him Out
01:39:13.989                             : en:Fire at Will: One
01:50:23.867                             : en:Copter Chase
01:52:28.658                             : en:"He's Pregnant"
02:00:36.771                             : en:Shaken, Not Stirred
02:05:56.882                             : en:Section Five

Of note is "Detected bit depth" (EAC3To's "VALID_BITS" tag), channel layouts despite the fact that it's an EAC3To FLAC file without a WAVEFORMATEXTENSIBLE_CHANNEL_MASK tag and all the tracks have bitrates, despite the fact that they're in an mkv and there are two variable bitrate tracks in there.

Subtitle track one has 2084 elements in it, subtitle two has only 36, that one must be forced subtitles, the last subtitle has 2460 elements, which is more than the first, so the first must be the regular subtitles and the last must be SDH.

Oh, right, the flac was muxed straight into the file using MKVMerge, so the UID is 3419832279421249968 and yet its statistics tags are still being applied to it correctly.

And then there's this:

General
Unique ID                                : 215636814642891718231343589359444197377 (0xA23A23CCC54ADBF1948103B2D1428001)
Complete name                            : D:\DTS-X.mka
Format                                   : Matroska
Format version                           : Version 4 / Version 2
File size                                : 123 MiB
Duration                                 : 3mn 32s
Overall bit rate                         : 4 861 Kbps
Encoded date                             : UTC 2015-11-15 15:35:39
Writing application                      : mkvmerge v8.5.1 ('Crosses') 64bit
Writing library                          : libebml v1.3.3 + libmatroska v1.4.4

Audio
ID                                       : 1
Format                                   : DTS
Format/Info                              : Digital Theater Systems
Format profile                           : X / MA / Core
Mode                                     : 16
Format settings, Endianness              : Big
Codec ID                                 : A_DTS
Duration                                 : 3mn 32s
Bit rate mode                            : Variable / Variable / Constant
Bit rate                                 : 4 859 Kbps / 4 859 Kbps / 1 509 Kbps
Channel(s)                               : Object Orientated / 8 channels / 6 channels
Channel positions                        : Object Orientated / Front: L C R, Side: L R, Back: L R, LFE / Front: L C R, Side: L R, LFE
Sampling rate                            :  / 48.0 KHz / 48.0 KHz
Bit depth                                :  / 24 bits / 24 bits
Compression mode                         :  / Lossless / Lossy
Stream size                              : 123 MiB (100%)
Language                                 : English
Default                                  : Yes
Forced                                   : No

DTS:X

and this:

General
Unique ID                                : 186823894588399821734555224452256288412 (0x8C8CF929A176D63082E6B973CE30CE9C)
Complete name                            : D:\DTS ES.mka
Format                                   : Matroska
Format version                           : Version 4 / Version 2
File size                                : 6.80 MiB
Duration                                 : 4s 11ms
Overall bit rate                         : 14.2 Mbps
Encoded date                             : UTC 2015-11-13 23:30:23
Writing application                      : mkvmerge v8.5.1 ('Crosses') 64bit
Writing library                          : libebml v1.3.3 + libmatroska v1.4.4

Audio #1
ID                                       : 1
Format                                   : DTS
Format/Info                              : Digital Theater Systems
Format profile                           : MA / ES Matrix / Core
Mode                                     : 16
Format settings, Endianness              : Big
Codec ID                                 : A_DTS
Duration                                 : 4s 10ms
Bit rate mode                            : Variable / Constant / Constant
Bit rate                                 : 2 812 Kbps / 1 509 Kbps / 1 509 Kbps
Channel(s)                               : 8 channels / 7 channels / 6 channels
Channel positions                        : Front: L C R, Side: L R, Back: L R, LFE / Front: L C R, Side: L R, Back: C, LFE / Front: L C R, Side: L R, LFE
Sampling rate                            : 48.0 KHz
Bit depth                                : 24 bits
Compression mode                         : Lossless / Lossy / Lossy
Stream size                              : 1.34 MiB (20%)
Language                                 : English
Default                                  : Yes
Forced                                   : No

Audio #2
ID                                       : 2
Format                                   : DTS
Format/Info                              : Digital Theater Systems
Format profile                           : MA / ES Discrete / Core
Mode                                     : 16
Format settings, Endianness              : Big
Codec ID                                 : A_DTS
Duration                                 : 4s 0ms
Bit rate mode                            : Variable / Constant / Constant
Bit rate                                 : 2 269 Kbps / 1 509 Kbps / 1 509 Kbps
Channel(s)                               : 7 channels / 7 channels / 6 channels
Channel positions                        : Front: L C R, Side: L R, Back: C, LFE / Front: L C R, Side: L R, Back: C, LFE / Front: L C R, Side: L R, LFE
Sampling rate                            : 48.0 KHz
Bit depth                                : 24 bits
Compression mode                         : Lossless / Lossy / Lossy
Stream size                              : 1.08 MiB (16%)
Language                                 : English
Default                                  : No
Forced                                   : No


Audio #3
ID                                       : 3
Format                                   : DTS
Format/Info                              : Digital Theater Systems
Format profile                           : ES Discrete / Core
Mode                                     : 16
Format settings, Endianness              : Big
Codec ID                                 : A_DTS
Duration                                 : 4s 0ms
Bit rate mode                            : Constant
Bit rate                                 : 1 509 Kbps
Channel(s)                               : 7 channels / 6 channels
Channel positions                        : Front: L C R, Side: L R, Back: C, LFE / Front: L C R, Side: L R, LFE
Sampling rate                            : 48.0 KHz
Bit depth                                : 24 bits
Compression mode                         : Lossy
Stream size                              : 737 KiB (11%)
Language                                 : English
Default                                  : No
Forced                                   : No

DTS ES in all it's glory.

[Frechdachs]

I have a question regarding a 2.0 24-bit DTS-HD MA file. I've decoded it with dcadec (eac3to v3.31) and the Arcsoft decoder (eac3to v3.28) and the file decoded with the Arcsoft decoder is quieter.

According to eac3to the DTS file has a Dialnorm value of -4dB. If my assumptions about Dialnorm are correct, that means that a gain has to be applied, since neither flac nor wav support Dialnorm. I thought that maybe the dialogue normalization wasn't taken into account while using the Arcsoft decoder, so I encoded it again with the "+4dB" option and tested the difference to the dcadec version. In order to do so I opened both audio streams in Reaper and aplied phase inversion to one of the streams. Now if they are identical they should cancel each other out resulting in complete silence. There was still a sound at about -120dB (deviation from the peak), which is absolutely inaudible, though.

Since I don't know much about Dialnorm I want to ask which version is correct, the one without the gain or the one with it and the dcadec version.
Also, is there a reason there is still a small measurable difference between the Arcsoft decoded file with the gain and the one decoded by dcadec, shouldn't they be the same if the gain is applied manually?
And finally, is there a better way to check if two audio files are identical?

On a side note, I did the same test with Reaper using a 5.1 24-bit DTS-HD MA track without a Dialnorm value. There was no measurable difference between Arcsoft and dcadec, meaning they are in fact completely identical.

[tebasuna51]

@Frechdachs

- Dialog Normalization was a interesting concept if all sounds respect it.
But there are many (TV advertisement, remastered CD's Loudness war, ...) than try to offer the max digital volume in the wrong concept than loud volume is better quality. Thats enforce to use the Volume knob between normalized and not normalized sounds.
eac3to is a transcoding tool and, by default, try, if can do, to not apply Dialog Normalization to offer the original sound without attenuation.
What is correct? Is your choice.

- There are other methods to compare wav's but for this pourpose your method is correct (I also use it).
Take in mind than apply -4dB gain (DialNorm), and after a manual +4dB gain, is not a lossless operation and your can lose a bit of precission.
The dcadec output is the lossless source.

[airsoft]

Originally Posted by Frechdachs

And finally, is there a better way to check if two audio files are identical?

E:\>md5sum Tomorrowland*.wav
8e8d58cc32c6432a0c5222c7086e6444 *Tomorrowland_ArcSoft_decoder_eac3to-3.31.wav
8e8d58cc32c6432a0c5222c7086e6444 *Tomorrowland_default_eac3to-3.31.wav

listing of cmd prompt command dir - filesize and name

8 989 397 060  Tomorrowland_ArcSoft_decoder_eac3to-3.31.wav
8 989 397 060  Tomorrowland_default_eac3to-3.31.wav

I hope it was helpful for you as technique. Seems both "libDcaDec" and "ArcSoft" produces identical files which means according my knowledge the original lossless audio

[AlexKane]

@Frechdachs

Assuming you want the best possible result (sorry), the correct "version" is the one produced by dcadec, since it does not apply 4db of gain reduction. As tebasuna51 mentioned, the Arcsoft file will be lower quality since the LSBs (the noisefloor @ 120db) get trashed as your source file is integer precision format.

[Megalith]

Is eac3to still unable to remove dialnorm from DTS-HD MA tracks?

[SeeMoreDigital]

Originally Posted by Megalith

Is eac3to still unable to remove dialnorm from DTS-HD MA tracks?

Come to think of it. I don't think I've ever seen my surround sound amplifier respond to a DTS-HD MA dialnorm flag. I've only ever seen it respond to a Dolby Digital dialnorm flag

[madshi]

The Dolby Digital spec requires Dolby Digital decoders to apply dialnorm by default, unless the user specifically disables it. DTS is not as stupid, fortunately, so DTS tracks have much less problems with that. When using libdcadec, dialnorm is by default ignored, so eac3to decoding no longer has any problems with that, anyway. eac3to still cannot fully remove dialnorm from DTS-HD tracks, though, because doing so would require to rewrite the whole HD frame structure, including CRCs etc, which is very complicated.

[Thunderbolt8]

does detecting DTS:X headphone tracks work differently from detecting norma DTS:X tracks? from what Ive heard using this info here https://github.com/foo86/dcadec/issues/37 does not lead to success in cases of headphone tracks.

maybe the dcadec developer could be able to figure something out as well in this case?

[Frechdachs]

Originally Posted by tebasuna51

- Dialog Normalization was a interesting concept if all sounds respect it.
But there are many (TV advertisement, remastered CD's Loudness war, ...) than try to offer the max digital volume in the wrong concept than loud volume is better quality.

But why is it used on Blu-rays? There are no advertisments interrupting the movie, so there should be no need to constantly have to adjust the volume.

Originally Posted by madshi

The Dolby Digital spec requires Dolby Digital decoders to apply dialnorm by default, unless the user specifically disables it. DTS is not as stupid, fortunately, so DTS tracks have much less problems with that. When using libdcadec, dialnorm is by default ignored, so eac3to decoding no longer has any problems with that, anyway.

That means I am misinterpreting dialnorm afterall. I thought that a dialnorm of -4dB would mean that in order to honor dialnorm a positive gain of 4dB has to be applied after decoding. But since you wrote that dialnorm is ignored for DTS-HD while using dcadec and not while using Arcsoft, there is rather a negative gain applied if dialnorm is honored (because in my test using dcadec resulted in a file that was 4dB louder than Arcsoft).

So if I get it straight this time, it meanst that when I use the Arcsoft decoder with the "+4dB" option, there is first a negative gain of -4dB applied (because of dialnorm) and then a positive gain again, which is retarded.

Is this correct?

Originally Posted by madshi

eac3to still cannot fully remove dialnorm from DTS-HD tracks, though, because doing so would require to rewrite the whole HD frame structure, including CRCs etc, which is very complicated.

You mean when input and output is both DTS? If encoding to flac for example, either dialnorm has to be ignored or there has to be a gain applied, because flac doesn't support dialnorm. Or does it?

[nevcairiel]

Originally Posted by Frechdachs

You mean when input and output is both DTS? If encoding to flac for example, either dialnorm has to be ignored or there has to be a gain applied, because flac doesn't support dialnorm. Or does it?

When eac3to decodes the DTS track to convert it to FLAC, dialnorm is simply ignored.

[AlexKane]

That means I am misinterpreting dialnorm afterall. I thought that a dialnorm of -4dB would mean that in order to honor dialnorm a positive gain of 4dB has to be applied after decoding.

You indeed misinterpret dialnorm, as it is the exact opposite of what you thought.
A dialnorm flag of -4db means that the decoder needs to apply 4db of gain reduction during decoding. In your case, Arcsoft respects the dialnorm flag while dcadec does not.
If all you want is "Extract the audio as it is stored" and encode it in another format, you need to use dcadec. You can then apply any level adjustment during playback.

[tebasuna51]

Originally Posted by Frechdachs

...
So if I get it straight this time, it meanst that when I use the Arcsoft decoder with the "+4dB" option, there is first a negative gain of -4dB applied (because of dialnorm) and then a positive gain again, which is retarded.

Is this correct?

Yes Arcsoft apply -4dB (dialnorm) and after eac3to apply +4dB.
But not retarded, when you recode the time to do the operations is not important. Is other problem:

You lose precission in round to int values.

For instance you have a sample with a int 24 bits volume value of 16777211.
When you apply -4dB you obtain a real value than you must round to int:

Int value     -4dB         Int Up   Int Down
---------  -------------  --------  --------
 16777211  10585704,5003  10585705  10585704

But round up or down whe you apply +4dB:

Int value     +4dB          Int
---------  -------------  --------
 10585705  16777211,7919  16777212
 10585704  16777210,2070  16777210

You never finish with the lossless value 16777211

[Frechdachs]

Thank you for all the replys. Everything makes sense now.

Originally Posted by AlexKane

You indeed misinterpret dialnorm, as it is the exact opposite of what you thought.

Yeah, I was probably misguided by the fact that the DTS track isn't very loud to begin with. It's a bit odd that the studio would want to reduce the volume even further during playback.

Originally Posted by tebasuna51

You lose precission in round to int values.
[...]
You never finish with the lossless value 16777211

That's what I meant when I said that it would be kind of stupid by me to do that. But thanks for the explaination on how the precision loss comes to be. Much appreciated.

[ndjamena]

https://github.com/MediaArea/MediaIn...4d5b9e34f94144

if (HD_SubStreams_Count == 4) Fill(Stream_Audio, 0, Audio_Codec, "ATMOS");

Is EAC3To's detection of ATMOS any more complicated than that?

Is there anything else I should look out for?

[kukushka]

eac3to + Nero Audio Decoder (Nero 7) are dropping first aac frame on decoding. always. along with faad. ffmpeg & Nero AAC Decoder 1.5.1.0 are doing it correctly. can be verified as example with extracted aac from qaac --no-delay encoding
possible solution - to feed first frame twice to decoder?
oh, and as i've heard, 5.1 aac decoding is broken in eac3to after 3.0.1 version, but it's probably nothing new with it
/just sayin'

[Music Fan]

Originally Posted by kukushka

eac3to + Nero Audio Decoder (Nero 7) are dropping first aac frame on decoding. always. along with faad. ffmpeg & Nero AAC Decoder 1.5.1.0 are doing it correctly.

Did you also try neroAacDec alone to see if the first aac frame is dropped ?