eac3to - audio conversion tool

[xkodi]

@xkodi

Arcsofrt EAc3 Deocdre is noz supported by eac3to...( or did i miss something?) so Nero is the best choice for eac3to DD+ decoding at the Moment.
Its better then libavcodec and its a Reference Decoder

you're completely correct.

[zeropc]

say, is there a good way to see what bit depth a tdh has? it would be helpful when you trancode to dts as it would save space. i mean why encode at 24 bit if the source is only 16 bt, right.

[tebasuna51]

say, is there a good way to see what bit depth a tdh has? it would be helpful when you trancode to dts as it would save space. i mean why encode at 24 bit if the source is only 16 bt, right.

Don't worry, dts don't have bitdepth only bitrate, select the desired bitrate: -768 or -1536 (default) for your space/quality requirements.

edit: If you have a DTS Master Audio encoder I think you need intermediate wav files and eac3to analyze the thd and output the appropriate bitdepth.

[odin24]

edit: If you have a DTS Master Audio encoder I think you need intermediate wav files and eac3to analyze the thd and output the appropriate bitdepth.

Is there a DTS Master audio encoder available... or just DTS?

[tebasuna51]

Is there a DTS Master audio encoder available... or just DTS?

Of course DTS-HD Master Audio Suite (http://www.dts.com/Pro-Audio_Software/DTS-HD_Master_Audio_Suite.aspx)

... or just DTS (http://www.minnetonkaaudioshop.eu/epages/61620177.sf)

[Snowknight26]

Whats the difference between a / and a . in the channels? For example:

4: AC3, English, 5.1 channels, 448kbps, 48khz, dialnorm: -27dB, -27ms
5: AC3, English, 3/1 channels, 448kbps, 48khz, dialnorm: -27dB, -27ms

[tebasuna51]

Whats the difference between a / and a . in the channels? For example:

5.1 or 3/2.1 is 3 front channels, 2 surround channels and 1 LFE channel

3/1 or 3/1.0 is 3 front channels and 1 surround channel, without LFE channel.

[rica]

Hi @madshi.
Splitted and remuxed back in two parts of one of my BD rips in m2ts with TSMuxer.
Demuxed audio and sup with eac3to.
Remuxed back to BD with TSMuxer.
Created UDF2.5 ISO image with ImageBurn.
Watched with Virtual Clone Drive.
The results:
You can watch sup with the first part while you can't do this with the second part.

This time i demuxed sup with TSMuxer itself and remuxed back to BD.
Created ISO.
Watched with Virtual Clone Drive.
The result; i can watch the subtitle on second part as well.

When i compared sups created by both programs i didn't see any difference?

Can you please check this?

[Ytterbium]

Hi, is there a way to stop the generation of a log file?

[Joniii]

I know eac3to automaticly detects if Blu-ray track needs channel mapping but could someone tell me which Blu-ray audio tracks need channel mapping, I can't remember if it was just PCM or was it needed for other formats too like TrueHD?

[ACrowley]

Of course DTS-HD Master Audio Suite (http://www.dts.com/Pro-Audio_Software/DTS-HD_Master_Audio_Suite.aspx)

... or just DTS (http://www.minnetonkaaudioshop.eu/epages/61620177.sf)


Na, i wouldnt by Surcode dts encoder. Its outdated . And you can buy the great new DTS Surround Audio Suite for the same Price(249$)
Surcode DVD DTS Pro cant do anything beyond 5.1 and the Quality is not so good

But the dts Support told me that when you have the discontinued DTS Pro Series ,therse no Reason to by the DTS Surround Audio Suite. It can do exactly the same,excpet the Streamplayer Decoder, with the same Output Quality. I have Pro Series ,but maybe i will by DTS Audio Suite soon.

[jj666]

Do you have anything to back up the claim on Surcode quality? Given the product is still being sold...

-jj-

[ACrowley]

Do you have anything to back up the claim on Surcode quality? Given the product is still being sold...

-jj-

The Optimal Encoder

[ACrowley]

Do you have anything to back up the claim on Surcode quality? Given the product is still being sold...

-jj-

The Optimal Encoder should be the Reference Encoder from DTS, no Doubt. As i say, you can get the new DTS Audio Surround Suite for the same Price ,249$

What i mean is that Surcode cant do DTS ES 6.1 /24/96.
So, theres no Point to by Surcode for 249$

[zeropc]

that's true about the current state for dts encoding software. but if you don't need to upgrade to the new suite, then there is no need. surcode does deliver high quality dts files just only up to 5.1. heck, if i could, i'd switch to the new one but $250 is just not in it these days ;-)

[rica]

Hi guys, whenever i try to enter the online shop, i get this:

Sorry, the DTS Online Store is closed for maintenance. We will re-open soon. Thank you!

[ACrowley]

that's true about the current state for dts encoding software. but if you don't need to upgrade to the new suite, then there is no need. surcode does deliver high quality dts files just only up to 5.1. heck, if i could, i'd switch to the new one but $250 is just not in it these days ;-)

When youve already dts pro series encoder ,theres no need to by dts audilo surround suite.
dts support told ,me that theres ne quality difference and you can do the same ,dts es 6.1 / 24/96, with both Encoders

But Surcode cant do dts es 6.1 ,24/96., that the Point :)
When youve already surcode and you dont need dts es 6.1, 24/96...no need to by/use dts pro series or the dts audio suite

What i mean is simple....when you plan to by a dts encoder, dont by surcode for 249$. Because you can get the dts surround audio suite for 249$ too, which can do everything except dts hd .It would be Nonsense!
Thats all what im talking about. However ,that the wrong Thread for those discussions

[jj666]

Ok, so the quality of Surcode output is fine, it's just lacking in certain features then.

-jj-

[zeropc]

that's correct jj

[odin24]

Guys, I didn't mean to hijack the thread... but good info about DTS and encoders.

I have a new question about eac3to. If I want to demux just one track (TrueHD) from a m2ts file would I use the following command.

eac3to <source m2ts directory> <TrueHD track #> <destination directory>audio.thd+ac3

I used this, and it worked, but I'm not sure if it demuxed, or converted :confused: I normally would user tsMuxeR for this but I really want to start using eac3to for everything.

Thanks,

[Greif]

What is the result if I demux a TrueHD track to:

- file.thd (is this the raw TrueHD?)
- file.thd+ac3 (??)
- file.ac3 (TrueHD transcoded to AC3?)

[rica]

What is the result if I demux a TrueHD track to:

- file.thd (is this the raw TrueHD?)
- file.thd+ac3 (??)
- file.ac3 (TrueHD transcoded to AC3?)

Blu-rays have to include an ac3 besides thd.
So extacting thd as alone doesn't mean anyhing since BD must have an additional ac3 as well for the backward compatibility.
Extracting ac3 does mean extracting separate ac3 without transcoding.
Extracting to thd+ac3 is needed for TSMuxer to recognize the file but TSMuxer is not enough by alone, second pass is needed by TSremux for trueHD:

http://forum.doom9.org/showthread.php?t=141125


_ _ _ _ _

[florinandrei]

Can't demux a file created by a Canon HF100 camcorder:

eac3to v2.68
command line: eac3to train.m2ts 1: video.mkv 2: audio.ac3
------------------------------------------------------------------------------
M2TS, 1 video track, 1 audio track, 0:00:25
1: h264/AVC, 1080i60 /1.001 (16:9)
2: AC3, 2.0 channels, 256kbps, 48khz, -67ms
[v01] Extracting video track number 1...
[a02] Extracting audio track number 2...
[a02] Applying (E-)AC3 delay...
[v01] Muxing video to Matroska...
Unfortunately the Haali Muxer cannot handle this source file.
It doesn't contain enough seek/recovery points.
Aborted at file position 32964608.

File is here, it's about 46 MB:

http://dl.getdropbox.com/u/29966/eac3to/train.m2ts

[yonta]

@florinandrei

eac3to train.m2ts 1: video.mkv -seekToIFrames 2: audio.ac3

[bmnot]

I wanted to try -analyzeBitdepth on this Dolby TrueHD demo for laughs but it fails saying:

[a02] Extracting audio track number 2...
[a02] Extracting TrueHD stream...
[a02] Decoding with libav/ffmpeg...
[a02] Remapping channels...
[libav] Substream 0 parity check failed
[libav] Substream 0 checksum failed
[libav] Substream 0 length mismatch.
[a02] The libav decoder reported an error while decoding.
Aborted at file position 16384.

here it is: http://www.mediafire.com/?giwnn4qgtm2

[nautilus7]

Is this an uncut file? The problem doesn't have to do with -analyzebitdepth. It seems damaged unless libav has a bug and can't decode it, while nero can.

[evdberg]

I used the -analyzebitdepth option and got this result:
[a02] Bit depth analyzation: max 24 bits, average 19 bits, most common 16 bits.

What does "most common 16 bits" exactly mean? Is the LSB zero or the MSB zero (or 0xFF if the value is negative)? Because it seems to be a 24 bits tracks, my assumption is that the MSB is zero. Am I right?
In other words: how does your analyzation exactly work?

[madshi]

Would it be possible to add something like "skip errors" to eac3to as well?
perhaps eac3to could indeed have a "continue on error" function (perhaps just for very specific classes of error like this one in particular) and flag the timecode of the error for later user verification?
The DTS-HD MA track was inside a M2TS file, cant use delaycut on a M2TS track, so a -SkipOnError function would be nice
This is on my to do list, but it's not as easy as it may sound. All the eac3to audio and video parsers currently depend on the bitstream being alright. If the video or audio bitstream is damaged that can result in weird problems of all kinds. Before I can even think about allowing to "ignore" errors in corrupted source files I'd first have to make all audio and video bitstream parsers more resistent against errors. E.g. the AC3 parser would have to be able to handle incorrect CRCs, missing sync words and all that stuff - and might even have to fix the stream, similar to what delaycut does. Now there are a dozen of different audio bitstream parsers in eac3to and 3 video bitstream parsers. So making all this more error tolerant will take quite a while...

got an error message regarding channel mixing:

Edit: v.2.68 didn't do the trick also....
Will check that out. Do you happen to have a small sample?

Hi @madshi.
Splitted and remuxed back in two parts of one of my BD rips in m2ts with TSMuxer.
Demuxed audio and sup with eac3to.
Remuxed back to BD with TSMuxer.
Created UDF2.5 ISO image with ImageBurn.
Watched with Virtual Clone Drive.
The results:
You can watch sup with the first part while you can't do this with the second part.

This time i demuxed sup with TSMuxer itself and remuxed back to BD.
Created ISO.
Watched with Virtual Clone Drive.
The result; i can watch the subtitle on second part as well.

When i compared sups created by both programs i didn't see any difference?

Can you please check this?
No. I'm not interested at all in why a specific combination of specific things you're doing with a number of applications does not work. If you find a clear bug in the sup file produced by eac3to, let me know. If not you'll have to find out yourself why that specific combination of things you're doing doesn't work...

Hi, is there a way to stop the generation of a log file?
No.

I know eac3to automaticly detects if Blu-ray track needs channel mapping but could someone tell me which Blu-ray audio tracks need channel mapping, I can't remember if it was just PCM or was it needed for other formats too like TrueHD?
Most tracks need channel remapping. The remapping which is needed even depends on the decoder which is used.

What is the result if I demux a TrueHD track to:

- file.thd (is this the raw TrueHD?)
- file.thd+ac3 (??)
- file.ac3 (TrueHD transcoded to AC3?)
HD DVD style TrueHD tracks consist of only TrueHD frames. These are identified by eac3to as being "TrueHD". The correct extension for such tracks is "*.thd". Blu-Ray style TrueHD tracks consist of TrueHD audio frames and AC3 audio frames. Both types of frames are interweaved. eac3to identifies these tracks as "TrueHD/AC3". The correct file extension is "*.thd+ac3". The file extension "*.ac3" is self explaining.

So, if you demux a HD DVD style TrueHD track to "*.thd", the track is demuxed as it is. If you demux it to "*.thd+ac3", eac3to creates AC3 frames on the fly and interweaves them with the HD DVD data to create a Blu-Ray compatible "TrueHD/AC3" track. If you demux a HD DVD style TrueHD track to "*.ac3", the TrueHD track is transcoded to AC3.

If you demux a Blu-Ray style TrueHD track to "*.thd", the track is demuxed with the interweaved AC3 frames stripped away. If you demux it to "*.thd+ac3", the track is demuxed unmodified. If you demux a Blu-Ray style TrueHD track to "*.ac3", the AC3 frames are demuxed as they are with the TrueHD frames stripped away.

It can be even more complicated: If you demux a Blu-Ray TrueHD/AC3 track by using e.g. "eac3to source 2: audio.thd+ac3 -192", eac3to strips the original interweaved AC3 frames away, creates new AC3 frames with 192kbit/s on the fly and interweaves them with the original TrueHD frames coming from the Blu-Ray track.

I hope you're finally confused now... :D

I wanted to try -analyzeBitdepth on this Dolby TrueHD demo for laughs but it fails saying:

[a02] The libav decoder reported an error while decoding.
Is eac3to able to decode this track? I think it's simply a broken/corrupted TrueHD track.

I used the -analyzebitdepth option and got this result:
[a02] Bit depth analyzation: max 24 bits, average 19 bits, most common 16 bits.

What does "most common 16 bits" exactly mean? Is the LSB zero or the MSB zero (or 0xFF if the value is negative)? Because it seems to be a 24 bits tracks, my assumption is that the MSB is zero. Am I right?
In other words: how does your analyzation exactly work?
When people talk about Blu-Ray audio bitdepth, they usually talk about 16bit vs. 24bit. However, some Blu-Ray PCM tracks are 20bit (stored as 24bit with the 4 LSBs being zeroed out). eac3to checks the real bitdepth of the audio track. That is it searches for the first non-zero bit, beginning with the least significant bit.

The report eac3to posted in your specific case means that some samples in the audio track have a full bitdepth of 24bit (meaning no wasted zeroed out bits), while most samples are limited to 16bit only (least 8 significant bits zeroed out). I've seen this from time to time: Specific scenes have 24bit, while other scenes have the 8 LSBs zeroed out. E.g. with POTC 1 only the Walt Disney intro is 24bit, the rest of the movie is 16bit. The "average" bitdepth reported by eac3to tells you the weighted average bitdepth of all audio samples.

[odin24]

It can be even more complicated: If you demux a Blu-Ray TrueHD/AC3 track by using e.g. "eac3to source 2: audio.thd+ac3 -192", eac3to strips the original interweaved AC3 frames away, creates new AC3 frames with 192kbit/s on the fly and interweaves them with the original TrueHD frames coming from the Blu-Ray track.

What would be the point of this, just to save space? Would there be noticable quality loss as opposed to keeping the interweaved AC3 track at 448 or 640 kb/s?

Thanks,

[rica]

No. I'm not interested at all in why a specific combination of specific things you're doing with a number of applications does not work. If you find a clear bug in the sup file produced by eac3to, let me know. If not you'll have to find out yourself why that specific combination of things you're doing doesn't work...


Thanks anyway.

[kurt]

Will check that out. Do you happen to have a small sample?


here you are: http://rapidshare.com/files/155218023/6.1.stream.dts (~10MB)

thx for looking into this :)

[madshi]

What would be the point of this, just to save space?
Of course to save space. If you know for sure that you will never need that interweaved AC3 frames but still want/need to get a Blu-Ray compatible TrueHD/AC3 track, using a low bitrate for the AC3 frames makes sense.

Would there be noticable quality loss as opposed to keeping the interweaved AC3 track at 448 or 640 kb/s?
If you ever play that interweaved AC3 track then of course. If you only ever play the interweaved TrueHD frames, then no.

[evdberg]

When people talk about Blu-Ray audio bitdepth, they usually talk about 16bit vs. 24bit. However, some Blu-Ray PCM tracks are 20bit (stored as 24bit with the 4 LSBs being zeroed out). eac3to checks the real bitdepth of the audio track. That is it searches for the first non-zero bit, beginning with the least significant bit.

The report eac3to posted in your specific case means that some samples in the audio track have a full bitdepth of 24bit (meaning no wasted zeroed out bits), while most samples are limited to 16bit only (least 8 significant bits zeroed out). I've seen this from time to time: Specific scenes have 24bit, while other scenes have the 8 LSBs zeroed out. E.g. with POTC 1 only the Walt Disney intro is 24bit, the rest of the movie is 16bit. The "average" bitdepth reported by eac3to tells you the weighted average bitdepth of all audio samples.

Thanks for your explanation. I have some additional questions though. You state you determine the bitdepth of every sample, but do you count this separately (int count[24]) or do you categorize a sample in the 16 or 24 bits domain and count the number of samples in each domain?
If it is the first case, do you take into account the chance that a higher bitdepth sample has its lower bits zeroed? For instance if you determine that the lower 4 bits are zero, there is a 1/16 chance that this still is a 24 bits sample.

[shambles]

The "average" bitdepth reported by eac3to tells you the weighted average bitdepth of all audio samples.

some hd dvds released by universal have truehd tracks where centre, left, right and lfe channels are constant 16bit but surrounds are constant 24bit.. so for tracks where different channels have different bitdepths perhaps it would make sense to print out separate info for each channel

[madshi]

Thanks for your explanation. I have some additional questions though. You state you determine the bitdepth of every sample, but do you count this separately (int count[24]) or do you categorize a sample in the 16 or 24 bits domain and count the number of samples in each domain?
If it is the first case, do you take into account the chance that a higher bitdepth sample has its lower bits zeroed? For instance if you determine that the lower 4 bits are zero, there is a 1/16 chance that this still is a 24 bits sample.
I count each bitdepth separately (int64 count[24]). E.g. eac3to reports the "The Fifth Element" English audio track to be "constant 20 bit". That is also exactly what a Sony employee has stated about that audio track. I'm aware that some samples have their lower bits at 0 even though the bits contain valid data. I'm compensating for that "somewhat" in my final statistical calculations. Probably it's not mathematically perfect, but it seems to do the job well enough...

some hd dvds released by universal have truehd tracks where centre, left, right and lfe channels are constant 16bit but surrounds are constant 24bit.. so for tracks where different channels have different bitdepths perhaps it would make sense to print out separate info for each channel
I was not aware of that different channels sometimes have different bitdepths. That's interesting. But still I'm not sure how complicated I should make the bitdepth analyzation report. I mean I could print out 10 pages of statistics and print out some funny graphs. But actually many people are not very interested in the statistics at all. So when I developed that eac3to bitdepth statistics report I limited the output so that it always fits into one single line...

[evdberg]

@madshi, personally I would appreciate some more indepth statistics for the analyzebitdepth option. 10 pages is definitely not necessary, but a few lines extra would be great. The post by shambles also explains the result I got, since I tested an Universal title.

[madshi]

Well, what kind of statistics do you want to have? The same I'm doing now, but for every channel (in case the channels are different)? Or what are you looking for?

[evdberg]

Yes, statistics per channel, but if possible only when there is a difference between the channels. Also I prefer a percentage instead of terms like 'most common' when the bitdepth is not constant.

[madshi]

Also I prefer a percentage instead of terms like 'most common' when the bitdepth is not constant.
You mean a list of bitdepths which occur, each with the percentage number? That could get rather ugly if the bitdepth is all over the place, don't you think?

[evdberg]

That could indeed get ugly, although I think that soundtracks are in essence 16, 20 or 24 bits. Maybe you can divide the bitdepths into those 3 groups?

[madshi]

Dolby recommends to those studios which used to use 16bit only, to up that to ca. 18bit. So I don't think we can limit bitdepth analyzation to only 16, 20 and 24.

[shambles]

perhaps the (possible) more detailed analysis could be placed in the log while keeping the main output neat and clean

[nautilus7]

I think the default bit analyzation should stay as it is and for more info the -analyzebitdepth could be changed to output more detailed reports. Like -logdts.

[Zwitterion]

I've got a feature request:
A switch that phase-shifts the surround channels by 90°. This is needed for correct DPL II downmixing using -down2 by files that have no phase-shifted rears (especially DTS).

example (AC3 sample (http://rapidshare.com/files/155535814/_VTS_01_1_T80_3_2ch_448Kbps_DELAY_0ms.ac3.html)):
http://img410.imageshack.us/my.php?image=down2sq0.png
When using just -down2 FR and RR cancel each other out.


Another idea:
DIRAC time stretching (http://www.dspdimension.com/technology-licensing/dirac/). The LE version is free to implement.
I don't know any free tool that does high-quality timestretching, so that would definitely be another killer feature and very useful for 23.976 --> 25 audio conversions.

[zeropc]

erm ... why does eac3to create a gap file, even calculates what is need to fix it but then it doesn't use it afterwards. you have to run a second run which adds a tremendous amount of time.

can't this be done in the first pass or is this absolutly nessecary?

[nlnl]

Vista 32 + eac3to 2.67 crashed remuxing PCM 5.1 24/96 from m2ts Chris Botti: Live with Orchestra and Special Guests to Flac format.
Please help!:helpful:

[menlvd]

Vista 32 + eac3to 2.67 crashed remuxing PCM 5.1 24/96 from m2ts Chris Botti: Live with Orchestra and Special Guests to Flac format.
Please help!:helpful:

plz read change log for 2.68 ver.
v2.68
* fixed crash when transcoding Blu-Ray/HD DVD track to FLAC

[madshi]

I've got a feature request:
A switch that phase-shifts the surround channels by 90°. This is needed for correct DPL II downmixing using -down2 by files that have no phase-shifted rears (especially DTS).
Would love to, but I don't know how to do it. Does anybody know any LGPL libraries or source code which can do high quality phase shifts? (if possible floating point in/out)

Another idea:
DIRAC time stretching (http://www.dspdimension.com/technology-licensing/dirac/). The LE version is free to implement.
I don't know any free tool that does high-quality timestretching, so that would definitely be another killer feature and very useful for 23.976 --> 25 audio conversions.
Looks good! Will gave this a try...

erm ... why does eac3to create a gap file, even calculates what is need to fix it but then it doesn't use it afterwards. you have to run a second run which adds a tremendous amount of time.

can't this be done in the first pass or is this absolutly nessecary?
It can't be done in the first pass. A 2nd pass is necessary. It's not done automatically because sometimes it might be preferred to not fix the detected gaps/overlaps.

[madshi]

got an error message regarding channel mixing
Fixed in v2.69.

@madshi, personally I would appreciate some more indepth statistics for the analyzebitdepth option.
Have added that to my to do list. Don't know when I will get to that, though...

[madshi]

eac3to v2.69 released

http://madshi.net/eac3to.zip

* added high precision SSRC resampler
* resampling "-quality" now allows "low", "high" (SSRC) or "ultra" (r8brain)
* resampling quality now defaults to "high" (SSRC)
* bitdepth is now analyzed separately for original vs. processed data
* fixed: downmixing 16 bit DTS tracks to 5.1 or 2.0 didn't work
* fixed: Sonic Decoder was incorrectly assumed to decode XXCh DTS files to 6.1
* for movies the Haali Muxer can't handle "-seekToIFrames" is suggested now
Please note that I'm not sure whether SSRC or r8brain is better for resampling. r8brain is a lot slower than SSRC, so hopefully it's a little bit better, but you be the judge. Because of the dramatic speed difference the highest quality SSRC mode is now the default resampling mode (also used for PAL speedup/slowdown). SSRC is somewhat limited in which conversions it likes to do exactly, though. Some sample rate conversions might be declined by SSRC. If you stumble over such a case, just use the "-quality=ultra" option to switch to r8brain instead.