Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Audio encoding

Closed Thread
 
Thread Tools Search this Thread Display Modes
Old 22nd December 2008, 09:16   #7481  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by krosswindz View Post
I have a DTS HD MA 6.1ch track @ 16 bit. When I try to extract the core or convert it to 5.1ch 1.5 MBps DTS eac3to seems to patch to 24bit DTS. Is there any way I can prevent it from patching it to 24bit and keep it at 16 bit. -16 switch works only with RAW/PCM audio I suppose.
Why does the patching bother you?

Quote:
Originally Posted by Chrishel View Post
Do you have any suggestions as to anything I could try to get eac3to to detect the dtswavs properly.
Yes. Make a sample available to me...

Quote:
Originally Posted by bmnot View Post
madshi, eac3to is hands down one of the best designed and supported pieces of software I've ever used. Is there anyway I can donate a little something for all the hard work you put into it?
Thanks! And no, right now there's no way to donate, cause such a donation would be difficult for me to handle tax wise. But I might change my mind next year...

Quote:
Originally Posted by williewonton View Post
While 2.84 isn't broken by the title "Game Plan" I've found that the audio isn't correctly sequenced. I'll do some more looking and see if I can make sense of it.
What do you mean exactly with "correctly sequenced"?
madshi is offline  
Old 22nd December 2008, 10:50   #7482  |  Link
Chrishel
Registered User
 
Join Date: Aug 2008
Posts: 3
Quote:
Originally Posted by madshi View Post
Yes. Make a sample available to me...
Check your PMs. :-)
Chrishel is offline  
Old 22nd December 2008, 11:56   #7483  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Chrishel View Post
Check your PMs. :-)
I'm checking my PMs, anyway, so there's no need to flood this thread with such posts.

-------

Your DTSWAV sample is strange. It begin with a lot of zero data and then half a DTS frame and then finally with the real DTS data. eac3to cannot automatically detect such a file as a valid DTSWAV file. Don't know, maybe the EAC extra features (which are very useful for ripping normal WAV data) are harmful for DTSWAV tracks? You could try playing around with the EAC options to see whether that makes any difference.

Furthermore the DTS data in the DTSWAV file is kind of broken. That's probably not the fault of EAC. The framesize is supposed to be 3585 bytes, but the data only contains 3584 bytes per frame. That's why eac3to didn't accept the DTS file extracted by DTSParser, either. I'll implement a workaround for such broken DTS files into the next eac3to build. So the next eac3to build will be able to detect the DTSParser extracted DTS file as:

Code:
DTS, 5.1 channels, 0:03:25, 20 bits, 1235kbps, 44.1khz
Also decoding will be possible with ArcSoft, libav and Nero, but not with Sonic.
madshi is offline  
Old 22nd December 2008, 13:00   #7484  |  Link
Chrishel
Registered User
 
Join Date: Aug 2008
Posts: 3
Thanks so much!
Chrishel is offline  
Old 22nd December 2008, 13:27   #7485  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 6,890
Quote:
Originally Posted by krosswindz View Post
I have a DTS HD MA 6.1ch track @ 16 bit. When I try to extract the core or convert it to 5.1ch 1.5 MBps DTS eac3to seems to patch to 24bit DTS. Is there any way I can prevent it from patching it to 24bit and keep it at 16 bit. -16 switch works only with RAW/PCM audio I suppose.
Maybe you are confused, only DTS MA (lossless) can have a exact bitdepth.

The core, or a downmix to 5.1 reencoded to a standard DTS (lossy), don't have any bitdepth, the samples are stored in frequency domain with a equivalent precission to 20-24 bits in uncompressed format, not matter if the source is 16 or 24 bits.

A -16 switch with a dts output (or other lossy format) have not sense.
__________________
BeHappy, AviSynth audio transcoder.

Last edited by tebasuna51; 22nd December 2008 at 13:32.
tebasuna51 is offline  
Old 22nd December 2008, 17:30   #7486  |  Link
krosswindz
Registered User
 
Join Date: Feb 2006
Location: kernel space
Posts: 245
Quote:
Originally Posted by madshi View Post
Why does the patching bother you?
If I can say bit depth is something equivalent to resolution of the sample stored. If the source is 16bit then while re-encoding if it is re-encoded as 24bit would this mean the encoder is filling up the missing data.

Quote:
Originally Posted by tebasuna51 View Post
Maybe you are confused, only DTS MA (lossless) can have a exact bitdepth.

The core, or a downmix to 5.1 reencoded to a standard DTS (lossy), don't have any bitdepth, the samples are stored in frequency domain with a equivalent precission to 20-24 bits in uncompressed format, not matter if the source is 16 or 24 bits.

A -16 switch with a dts output (or other lossy format) have not sense.
Very true I am confused on this.

What I am trying to do is to keep everything at the same except downmix to 5.1ch keeping every other same as possible.

edit: @madshi it would be great if there was an option to prevent patching and let the default be that the audio be patched.

Last edited by krosswindz; 22nd December 2008 at 17:43.
krosswindz is offline  
Old 22nd December 2008, 18:30   #7487  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 6,890
Quote:
Originally Posted by krosswindz View Post
If I can say bit depth is something equivalent to resolution of the sample stored. If the source is 16bit then while re-encoding if it is re-encoded as 24bit would this mean the encoder is filling up the missing data.
Still confused. After a lossy encode forget the source resolution, if we can recover the source resolution (even if is poor) we have a lossless encoder.

For a lossy output the best procedure is, like madshi know, decode to high resolution (64 or 32 bit float) make functions (mix, speed, resample, ..) with max resolution and, at end, down to the max resolution supported by the encoder (24 bit int for Surcode DTS, 32 bit float for NeroAacEnc, ...).

In lossless formats the precision is know by the bitdepth, in lossy formats by the bitrate (forget the bitdepth).
__________________
BeHappy, AviSynth audio transcoder.
tebasuna51 is offline  
Old 22nd December 2008, 19:10   #7488  |  Link
Thunderbolt8
Registered User
 
Join Date: Sep 2006
Posts: 2,197
its basically that the sound quality in it remains at 16-bit, but it seems like the way it is stored can only be 24-bit in that case. so theres no downmixing or artificial upgrading of the original quality, it remains untouched, its only the 'package' which transports that sound content which gets changed. but apparently it cannot be made another way.

(correct me if im wrong)
Thunderbolt8 is offline  
Old 22nd December 2008, 20:03   #7489  |  Link
williewonton
Registered User
 
Join Date: Dec 2008
Posts: 10
madshi

The video runs apparently OK, but the audio track repeats the opening theme over and over and doesn't progress to the dialog etc. So it is an odd one. In a seamless branching title, what is the maximum you allow for/handle?
williewonton is offline  
Old 22nd December 2008, 23:48   #7490  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by krosswindz View Post
If I can say bit depth is something equivalent to resolution of the sample stored.
If you're talking about a LPCM or WAV track then you're right. If you're talking about a lossy DTS track then you're flat out wrong. As tebasuna51 already explained, DTS tracks do not store samples in any specific bitdepth. The DTS "bitdepth" header field only has pure informational character and tells us which bitdepth the original PCM audio master had which was fed into the DTS encoder. But the DTS track itself is not bound to any specific bitdepth.

Quote:
Originally Posted by krosswindz View Post
@madshi it would be great if there was an option to prevent patching
No, because IMO such an option would be totally useless. If you find a technically correct argument for adding such an option, then please let me know. But right now I don't see any such argument...

Quote:
Originally Posted by williewonton View Post
The video runs apparently OK, but the audio track repeats the opening theme over and over and doesn't progress to the dialog etc. So it is an odd one. In a seamless branching title, what is the maximum you allow for/handle?
There's no max. You can have thousands of m2ts parts, no problem, just takes longer.

I remember that there was a Blu-Ray disc which really had a broken audio track just like you describe it. Are you sure that eac3to borked up this audio track? I rather guess that the audio data in the m2ts files is broken. However, IIRC there's another audio track which is working fine. So please check all audio tracks. You may find one which is correctly working...
madshi is offline  
Old 23rd December 2008, 01:30   #7491  |  Link
krosswindz
Registered User
 
Join Date: Feb 2006
Location: kernel space
Posts: 245
@madshi/tebasuna51 thanks for the explanation, I didnt know that the DTS "bitdepth" header field was only for informational purpose. I know have a better understanding of what you mean.

@madshi: Really love your tool. Is there any future plan for supporting DTS pro series encoders?
krosswindz is offline  
Old 23rd December 2008, 01:43   #7492  |  Link
rica
Registered User
 
Join Date: Mar 2008
Posts: 2,021
Quote:
Originally Posted by Atak_Snajpera View Post
Transcoding from one lossy format to another always means loss in quality.
Not always, this is a myth which everybody believes in.

It is up to the capabilities of mostly decoder, plus encoder.

I have lots of dts samples sound better than the original ac3s.

Edit: Before getting the objections, i should say:
Why do you use an Hi-End CD Player while you can listen the same CD on an ordinary player?
Because of its HW decoder (plus of its DAC); correct?
So if you can match the Hi-Fi decoder filter with an appropriate encoder filter; why not?
This is why i said "decoders are more important than the encoders in transcoding" before...
And madshi knows what they are and selected those Hi-Fi decoders as default in his tool...
So extracting wavs using an Hi-Fi decoder and re-encoding them with an external Pro encoder would be the best choice.
_ _ _ _ _ _

Last edited by rica; 23rd December 2008 at 03:29.
rica is offline  
Old 23rd December 2008, 03:26   #7493  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 6,890
Quote:
Originally Posted by rica View Post
Not always, this is a myth which everybody believes in.
Well you can filter noise or modify anything wrong in the original. But, with a correct source, isn't a myth.
Quote:
It is up to the capabilities of mostly decoder, plus encoder.
No.
Quote:
I have lots of dts samples sound better than the original ac3s.
Please check your players settings.
Quote:
"decoders are more important than the encoders in transcoding"
No.
__________________
BeHappy, AviSynth audio transcoder.
tebasuna51 is offline  
Old 23rd December 2008, 03:35   #7494  |  Link
rica
Registered User
 
Join Date: Mar 2008
Posts: 2,021
"No" would not explain anything.
Except telling the same story...
rica is offline  
Old 23rd December 2008, 04:48   #7495  |  Link
alc0re
Registered User
 
Join Date: Jun 2008
Posts: 91
Sorry if this has been asked before or if it is a stupid question :

I have been using eac3to to extract video/audio/subs from bluray disks.

I noticed something though and I have a question.

What exactly does the removal of dialog normalization do?

I ask because when I extract an ac3 (dolby digital) track with dialog normalization, and let eac3to remove the dialog normalization, when I play my final encoded avchd structured dvd9 in either my bluray player or my PS3 the audio is really really low and I have to crank up my receiver's volume almost all the way up to hear anything. If I fast forward at like 1.5x realtime where you can still hear the audio its normal level, but as soon as you hit play again it drops the audio down again.

I blamed a PS3 firmware update for awhile but I exchanged my PS3 for a panasonic DMP-BD35K bluray player and it has the same behavior.

I finally got around to trying the -keepDialnorm and now I can hear the audio with my receiver's volume at a normal range. What exactly is removing the dialog normalization buying me and why is it not recommended to use the -keepDialnorm switch? Is there a way to extract the ac3 audio that already has had its dialog normalization removed from my dvd9s and do the reverse process that removing the dialog normalization does? Anyone else have/seen this issue? Perhaps its my receiver? (Yamaha HTR-5940)

Last edited by alc0re; 23rd December 2008 at 05:43.
alc0re is offline  
Old 23rd December 2008, 10:24   #7496  |  Link
Mercury_22
Registered User
 
Join Date: Dec 2007
Posts: 1,138
Sorry for n00b question but How can I demux to specific location using "-demux " command ?

__________________
Intel UHD Graphics 750; Win 10 22H2
Mercury_22 is offline  
Old 23rd December 2008, 10:30   #7497  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by krosswindz View Post
Is there any future plan for supporting DTS pro series encoders?
There is AGM output support (for older DTS pro encoders). Don't know if anything more than that will come.

Quote:
Originally Posted by rica View Post
Not always, this is a myth which everybody believes in.
No, it is not a myth. It's a technical fact that every encoding process to a lossy format throws away information. There may be situations where transcoding from lossy to lossy can result in an improvement. E.g. (as tebasuna51 said) if you post process the audio data in a good way. Or e.g. if one of the decoders in your receiver is broken. But these are really exceptions. It's a fact that every straight transcoding from lossy to lossy format loses audio information.

Quote:
Originally Posted by alc0re View Post
What exactly does the removal of dialog normalization do?
Dialnorm can be set to any value between 0 and 31. According to the AC3 specification both 0 and 31 means: No dialnorm processing. Now any dialnorm processing *lowers* the volume of the audio track. That means removing the dialnorm (which is what eac3to is doing) should result in *higher* volume. Currently eac3to sets dialnorm to 0. Unfortunetely dialnorm set to 1 means lowering volume a lot. So incorrectly working decoders might think that a dialnorm value of 0 means even lower volume than dialnorm 1. But the documentation clearly states that a dialnorm value of 0 shall be treated as "no dialnorm processing" (which means max volume). And all the PC AC3 decoders correctly see value 0 as "dialnorm processing deactivated".

Quote:
Originally Posted by alc0re View Post
I ask because when I extract an ac3 (dolby digital) track with dialog normalization, and let eac3to remove the dialog normalization, when I play my final encoded avchd structured dvd9 in either my bluray player or my PS3 the audio is really really low and I have to crank up my receiver's volume almost all the way up to hear anything. If I fast forward at like 1.5x realtime where you can still hear the audio its normal level, but as soon as you hit play again it drops the audio down again.

I blamed a PS3 firmware update for awhile but I exchanged my PS3 for a panasonic DMP-BD35K bluray player and it has the same behavior.
IMHO the decoders in the PS3 and Panasonic are not working correctly. Or maybe my AC3 specification is outdated? Anyway, the documentation clearly says that dialnorm 0 is "reserved". So I think it's not really good that eac3to uses it. That means I'll change it to 31 in the next build. I think that should fix the problem you're seeing. However, I believe to remember that some Sony Blu-Rays had a dialnorm value of 0, too. Well, anyway...
madshi is offline  
Old 23rd December 2008, 10:32   #7498  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Mercury_22 View Post
Sorry for n00b question but How can I demux to specific location using "-demux " command ?
By changing the command prompt "current directory" to the wanted location. (e.g. "d:", followed by "cd movies").
madshi is offline  
Old 23rd December 2008, 10:42   #7499  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by DrNein View Post
When demuxing an AVC TS, the resolution changes from broadcast standard 1088i to 1084i. Why is that? Is there a way to leave it alone or else change to 1080?
Quote:
Originally Posted by madshi
I've done several checks:

(1) MediaInfo reports 1920x1088 for the original video track. And it reports 1920x1080 for the eac3to processed track.

(2) When playing back the original TS file with the Cyberlink h264 decoder, the video renderer gets 1920x1088 pixels. When playing back the eac3to processed video track with the same decoder, the video renderer gets 1920x1080 pixels.

(3) Same as (2), but with Sonic h264 decoder.

(4) Same as (2), but with ffmpeg/libav h264 decoder.

I think that are enough proofs that eac3to's processing is 100% correct. The cropping down to 1080 from 1088 is a (surprisingly) complicated calculation and tsMuxeR evidently does it in the wrong way. It's so complicated because the cropping size must be multiplied several times, depending on some conditions. E.g. with your sample, eac3to writes a cropping value of "2" into the video bitstream. This must be multipled by 2 twice to get to the correct cropping size of "8". tsMuxeR seems to forget one of the multiplications, that's why it incorrectly reports 1084i instead of 1080i.

Short summary: This is a(nother) bug in tsMuxeR.
Quote:
Originally Posted by DrNein
Well, I suppose that is sort of good news then

Hopefully, tsMuxeR will be corrected for this and a few other things soon (such as 25.01 FPS output and THD handling). In the meantime, do you think it is safe to use tsMuxeR to mux such files? That is, is it only reporting wrong but not altering them?
I don't know. Why do people keep asking me questions about tsMuxeR?

Quote:
Originally Posted by DrNein
Do you plan to add TS and/or M2TS output to eac3to?
No.

Quote:
Originally Posted by DrNein
Also, I have not really noticed any problem with playback of most most 1088 AVC files however 1088 MPEG-2 with DXVA often display a grey bar in MPC-HC but not in PowerDVD. Can eac3to correct those MPEG-2?
No. In MPEG2 it's not as easy as in h264.

Quote:
Originally Posted by DrNein
Is there any potential drawback to eac3to automatically processing to 1080 (with either codec)?
I see none. Actually I think it has several advantages. E.g. with the original file the video renderer actually gets 1088 lines. What will the renderer do with that? Display all of them? Then it must downscale the image, which means we don't have 1:1 pixel mapping, anymore. With the eac3to cropped video bitstream the video renderer is only getting 1080 lines, so we get perfect 1:1 pixel mapping. You can see that for yourself: If you play the original TS file, you'll see some garbage lines at the bottom of the screen. With the eac3to processed stream these garbage lines are gone.
madshi is offline  
Old 23rd December 2008, 11:10   #7500  |  Link
Mercury_22
Registered User
 
Join Date: Dec 2007
Posts: 1,138
Quote:
Originally Posted by madshi View Post
By changing the command prompt "current directory" to the wanted location. (e.g. "d:", followed by "cd movies").
Thanks !
__________________
Intel UHD Graphics 750; Win 10 22H2
Mercury_22 is offline  
Closed Thread

Tags
eac3to

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 08:32.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.