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

Reply
 
Thread Tools Search this Thread Display Modes
Old 8th November 2009, 18:51   #1021  |  Link
nurbs
Registered User
 
Join Date: Dec 2005
Posts: 1,460
I don't know about audio, but if you use a lossy source with video codecs then codecs that are similar to the codec the source has been encoded with have an advantage.
nurbs is offline   Reply With Quote
Old 8th November 2009, 19:23   #1022  |  Link
Abradoks
Registered User
 
Join Date: Mar 2008
Posts: 71
Quote:
Originally Posted by nurbs View Post
I don't know about audio, but if you use a lossy source with video codecs then codecs that are similar to the codec the source has been encoded with have an advantage.
Right. But when somebody makes a comparison between 20 Mbit/s BD source and 3 Mbit/s rips made with different codecs nobody blames him (even if he uses non-DCT codec).
Abradoks is offline   Reply With Quote
Old 8th November 2009, 20:19   #1023  |  Link
yesgrey
Registered User
 
Join Date: Sep 2004
Posts: 1,295
Abradoks,
I've sent you a PM with a link where you can get a sample of a lossless 5.1 48kHz 24bit audio file. If you could rerun the test with it it would be great...

Thanks.
yesgrey is offline   Reply With Quote
Old 8th November 2009, 21:46   #1024  |  Link
honai
Guest
 
Posts: n/a
The test was biased because inherently the "source" DTS has already been processed/filtered in favor of DTS.

Also, since we know that the LPCM track of At World's End is 16 bit, there is a very high probability that the source for the DTS was also 16 bit, regardless of the DTS header information.
  Reply With Quote
Old 8th November 2009, 21:53   #1025  |  Link
Abradoks
Registered User
 
Join Date: Mar 2008
Posts: 71
Quote:
Originally Posted by yesgrey3 View Post
I've sent you a PM with a link where you can get a sample of a lossless 5.1 48kHz 24bit audio file. If you could rerun the test with it it would be great...
Code:
Name		|  SNR	| PEAQ		||  SNR	  | PEAQ
		448 kbps		224 kbps
DD6, L		| 22.618| -0.205	||  16.188| -2.471
aften, L	| 18.541| -0.523	||  12.844| -3.115
DD6, C		| 23.644| -0.175	||  16.875| -2.203
aften, C	| 18.571| -0.377	||  12.790| -3.109
Quote:
Originally Posted by honai View Post
The test was biased because inherently the "source" DTS has already been processed/filtered in favor of DTS.

Also, since we know that the LPCM track of At World's End is 16 bit, there is a very high probability that the source for the DTS was also 16 bit, regardless of the DTS header information.
I know. But as I've already mentioned it doesn't make a difference. Look at results measured on the 24-bit PCM provided by yesgrey3.
Abradoks is offline   Reply With Quote
Old 8th November 2009, 22:16   #1026  |  Link
raquete
Registered User
 
Join Date: Nov 2004
Location: Brazil
Posts: 745
Quote:
Originally Posted by Abradoks View Post
I know. But as I've already mentioned it doesn't make a difference. Look at results measured on the 24-bit PCM provided by yesgrey3.
i can't post in mumbers how much tests i did and i trust in your results but in particular, i'm interested to hear as encoding with aften for long time feel better results when comparing with softencode or with the Dolby encoder inside sound forge 9.

you and yesgrey3 could post yours samples for members apreciations if don't break forum's rules.

regards!
raquete is offline   Reply With Quote
Old 8th November 2009, 23:39   #1027  |  Link
yesgrey
Registered User
 
Join Date: Sep 2004
Posts: 1,295
Quote:
Originally Posted by Abradoks View Post
But as I've already mentioned it doesn't make a difference. Look at results measured on the 24-bit PCM provided by yesgrey3.
Yes, DD still has better numbers, but the difference is smaller.
I would also like to see the numbers with 640kbps...

Quote:
Originally Posted by raquete View Post
you and yesgrey3 could post yours samples for members apreciations if don't break forum's rules.
I have considered posting the file here, but decided not to do it because I did not knew if I could be violating any rule...
I've extracted the period between minute 1:00 to 2:10 of "quantum of solace" Blu-ray.
yesgrey is offline   Reply With Quote
Old 9th November 2009, 02:53   #1028  |  Link
SquallMX
Special SeeD
 
Join Date: Nov 2002
Location: Mexico
Posts: 333
Quote:
Originally Posted by jruggle View Post
That is actually very good to know. I don't have access to a DD encoder. I would not be opposed to changing the defaults if that's what users would prefer. What are the other cutoffs for different bit rates and channel layouts?

edit: also, could you provide some small snippets of the DD-encoded files so that I can see what encoding features and params they use to get the better PEAQ and SNR values?
Aften use by default the average frequency cut of Dolby Encoders, example:

Dolby 5.1 AC3 448 Kbps, Freq 14/20, Aften 17 Khz
Dolby 2.0 AC3 192 Kbps, Freq 10/20, Aften 15 Khz

Probably because the lack of channel/frequency coupling in Aften, if I recall correctly you need 572/256 Kbps for true 20 Khz cut off on 5.1/2.0 audio files encoded by dolby encoders, ironical you can't get more than that in an official encoder, but with aften you can go up to 24 Khz if you want to.

SquallMX is offline   Reply With Quote
Old 10th November 2009, 12:16   #1029  |  Link
raquete
Registered User
 
Join Date: Nov 2004
Location: Brazil
Posts: 745
Quote:
Originally Posted by yesgrey3 View Post
I have considered posting the file here, but decided not to do it because I did not knew if I could be violating any rule...
I've extracted the period between minute 1:00 to 2:10 of "quantum of solace" Blu-ray.
i posted few samples in lots of formats with round 10 seconds in the forum some time ago in differents threads and i don't break rules as the idea is too share and change informations then, i can't see any problem.
raquete is offline   Reply With Quote
Old 10th November 2009, 14:31   #1030  |  Link
Abradoks
Registered User
 
Join Date: Mar 2008
Posts: 71
Quote:
Originally Posted by yesgrey3 View Post
Yes, DD still has better numbers, but the difference is smaller.
It's not because of lossless source. With 1 minute sample I've previously posted I get results that differ from the whole track results in the same way:
Code:
Name		|  SNR	| PEAQ
		448 kbps
DD6, L		| 23.382| -0.205
aften, L	| 19.154| -0.582
And the difference is not "smaller". Look at PEAQ values.
Quote:
Originally Posted by yesgrey3 View Post
I would also like to see the numbers with 640kbps...
Code:
Name		|  SNR	| PEAQ
		640 kbps
DD6, L		| 27.809| -0.039
aften, L	| 22.764| -0.172
DD6, C		| 28.911|  0.003
aften, C	| 22.982| -0.056
Quote:
Originally Posted by raquete View Post
i'm interested to hear as encoding with aften for long time feel better results when comparing with softencode or with the Dolby encoder inside sound forge 9.
Here is "Quantum of Solace" source wav, output.448.DD6.ac3 and output.224.DD6.ac3. You can get compiled aften from MeGUI server. Try to compare DD encoding with default aften frequency range and with tweaked through -w option.
But you should note, that all these numbers are mostly about development. If you feel that aften "sounds better" then you should use it.
Abradoks is offline   Reply With Quote
Old 10th November 2009, 18:44   #1031  |  Link
shon3i
BluRay Maniac
 
shon3i's Avatar
 
Join Date: Dec 2005
Posts: 2,419
Quote:
Originally Posted by Abradoks
Look at PEAQ values.
And you start to use ears instead some numbers. AC3 use different acoustic models to cheat human perception, that mean if some number show that codec A is better than B, B still can be fully transparent to A. Same as PSNR and SSIM for Video measuring.
shon3i is offline   Reply With Quote
Old 10th November 2009, 21:07   #1032  |  Link
yesgrey
Registered User
 
Join Date: Sep 2004
Posts: 1,295
Quote:
Originally Posted by Abradoks View Post
And the difference is not "smaller". Look at PEAQ values.
Yes, I forgot to look at those...

Quote:
Originally Posted by Abradoks View Post
Code:
Name		|  SNR	| PEAQ
		640 kbps
DD6, L		| 27.809| -0.039
aften, L	| 22.764| -0.172
DD6, C		| 28.911|  0.003
aften, C	| 22.982| -0.056
I know that maybe it could be slightlt off-topic, but could you, by any chance, also show the numbers for DTS 1510kbps?
It would be interesting to compare the results with the AC3 640kbps, since some tests showed that AC3 640kbps is superior to DTS 1510kbps...
yesgrey is offline   Reply With Quote
Old 11th November 2009, 20:22   #1033  |  Link
raquete
Registered User
 
Join Date: Nov 2004
Location: Brazil
Posts: 745
Abradoks
first i thank you to send the samples and source.

to encode with aften i use WAVtoAC3Enc4.1, all adjusts default(aften R832, newers versions don't work in my old win2000).

224k: sounds equal, frequences don't go ahead 9KHz.

448: WOW, your sample sounds good, is clever and go ahead 20KHz.
with aften sounds good too but don't go ahead 17KHz.

see?!? hearing we can feel differences!

...now i have requests to use it in WAVtoAC3Enc

cheers and thanks again.

Last edited by raquete; 11th November 2009 at 20:27.
raquete is offline   Reply With Quote
Old 11th November 2009, 21:16   #1034  |  Link
Abradoks
Registered User
 
Join Date: Mar 2008
Posts: 71
Quote:
Originally Posted by raquete View Post
with aften sounds good too but don't go ahead 17KHz
Please try to compare aften encodings with different frequency cutoffs. Here are six samples: aften_wide ("-w 60" for 448 kbit/s and 224 kbit/s; full range), aften_spec ("-w 48" and "-w 8"; DD range) and aften ("-w 34" and "-w 6"; default range).
Abradoks is offline   Reply With Quote
Old 12th November 2009, 15:58   #1035  |  Link
raquete
Registered User
 
Join Date: Nov 2004
Location: Brazil
Posts: 745
Abradoks, i'm horrible in command lines and i use WAVtoAC3Enc4.1(Aften) and don't know how to do the rights adjusts in this prog too.
can you write the command lines to use Aften with yours parameters posted please?

thanks.
raquete is offline   Reply With Quote
Old 12th November 2009, 17:34   #1036  |  Link
arestarh
Registered User
 
arestarh's Avatar
 
Join Date: Nov 2008
Location: Ukraine
Posts: 34
Quote:
Originally Posted by raquete View Post
Abradoks, i'm horrible in command lines and i use WAVtoAC3Enc4.1(Aften) and don't know how to do the rights adjusts in this prog too.
can you write the command lines to use Aften with yours parameters posted please?

thanks.
Maybe this option in WAVtoAC3Enc is equel to key -w in command line mode:
arestarh is offline   Reply With Quote
Old 12th November 2009, 19:26   #1037  |  Link
Abradoks
Registered User
 
Join Date: Mar 2008
Posts: 71
Quote:
Originally Posted by raquete View Post
Abradoks, i'm horrible in command lines and i use WAVtoAC3Enc4.1(Aften) and don't know how to do the rights adjusts in this prog too.
can you write the command lines to use Aften with yours parameters posted please?
It was just a description of samples I was linking to. You can just download it, no need to encode by yourself.

Last edited by Abradoks; 12th November 2009 at 23:52.
Abradoks is offline   Reply With Quote
Old 13th November 2009, 13:49   #1038  |  Link
raquete
Registered User
 
Join Date: Nov 2004
Location: Brazil
Posts: 745
Quote:
Originally Posted by Abradoks View Post
It was just a description of samples I was linking to. You can just download it, no need to encode by yourself.
ok, i got the files to hear and measure.

output.224.aften don't go ahead 9KHz.
output.224.aften_spec don't go ahead 9,5KHz.
output.224.aften_wide go ahead 23KHz.
output.448.aften don't go ahead 17KHz.
output.448.aften_spec go ahead 20,5KHz.
output.448.aften_wide go ahead 23KHz.

"wide" samples are very very good, 224.aften_wide and 448.aften_spec is very nice too!

i could feel better differences if was know musics but the samples are clevers.

thanks and regards!
raquete is offline   Reply With Quote
Old 15th November 2009, 19:36   #1039  |  Link
jruggle
Registered User
 
Join Date: Jul 2006
Posts: 276
Quote:
Originally Posted by Abradoks View Post
Try to compare DD encoding with default aften frequency range and with tweaked through -w option.
But you should note, that all these numbers are mostly about development. If you feel that aften "sounds better" then you should use it.
So far the only significant difference I've found between the Aften output and the DD output is channel coupling. The bandwidth reported for DD is the upper end of the coupling range. Aften does not support channel coupling yet.

The problem I've had implementing it in Aften has been the complexity of the code due to simd and multi-threading. As an experiment, I stripped all of that out and was able to get working channel coupling (although it was much slower obviously). I'm still working on trying to get it ported over to Aften.

Based on the AC-3 features used, the only other place where there could be some differences is in the exponent strategy decision. I have not analyzed this very thoroughly yet. I think Aften's exponent strategy implementation is pretty good though.

edit: oh, and not that it will matter a lot, but for any further comparisons you should turn on block switching in Aften since DD does block switching by default.

Last edited by jruggle; 15th November 2009 at 19:38.
jruggle is offline   Reply With Quote
Old 15th November 2009, 19:50   #1040  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 6,890
Quote:
Originally Posted by jruggle View Post
...
As an experiment, I stripped all of that out and was able to get working channel coupling (although it was much slower obviously). I'm still working on trying to get it ported over to Aften.
...
Very much appreciated your effort to improve Aften.
__________________
BeHappy, AviSynth audio transcoder.
tebasuna51 is offline   Reply With Quote
Reply

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 23:22.


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