IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
AC3 5.1 to AAC 5.1 Suitable Mode/Profile/Bitrate?
Makaveli7184
post Jun 21 2012, 15:25
Post #1





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



I have a large number of 5.1 AC3 audio file encoded at CBR 384 kbps that i want to transcode to AAC. Channel mapping is not a problem. Bytes are scarce, so my goal is to minimize bitrate while preserving audio transparency.
Now i've read in various online posts that AAC standards recommend around 260 kbps for 5.1 audio. I'm using a certain application (i'm not interested in suggestions of using another one or another encoder for that matter) that uses up-to-date AAC codecs. This application gives me the option of encoding in CBR AAC (bitrate ranging from 4 to 448 kbps in increments of 4, and profiles HE2/HE/LC are completely user selected) and VBR AAC (where i can choose a quality preset from 1 to 5, 5 being highest quality and largest filesize and where profile is strictly automatic, HE-AAC 2 kicks in when using Q1, HE-AAC when using Q2, and from Q3 to Q5, format profile is LC).

Giving all the above, and knowing that audio transparency and quality is usually subjective, what would be a good option?
My initial preference is using VBR Q3/LC which gives me 5.1 audio files encoded at around 250 ~ 300 kbps. Do i need to go higher?? Can i go lower without losing perceptive quality?? Should i consider lower CBR with HE profile??

Waiting for some suggestions. Thank you.
Go to the top of the page
+Quote Post
probedb
post Jun 21 2012, 16:02
Post #2





Group: Members
Posts: 1232
Joined: 6-September 04
Member No.: 16817



Do some tests and find out. You're going from lossy to lossy anyways so there's some loss whatever you do but whether or not you will notice is down for you to decide smile.gif
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 21 2012, 16:39
Post #3





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



Well, i appreciate your input, but that's not what i'm looking for. I'm looking for external input or suggestions, regardless of what my perception of quality is. It doesn't have to be exact science, but rather some simple pointers or some "this is probably what i would've done"s.

This post has been edited by Makaveli7184: Jun 21 2012, 16:52
Go to the top of the page
+Quote Post
onkl
post Jun 21 2012, 22:00
Post #4





Group: Members
Posts: 125
Joined: 27-February 09
From: Germany
Member No.: 67444



5.1 AC3 with only 384 kbps is already quite low. If you want to reduce bitrates further, I'd suggest to downsample to stereo.

This post has been edited by onkl: Jun 21 2012, 22:03
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 25 2012, 20:29
Post #5





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



Well, let's say that i wanted to convert an original AC3 to another AC3 file (audio unedited), would there have been any need to go higher then the original 384 kbps of the original AC3??
I ask because various listening tests have proven (albeit many of them indirectly) that AAC 5.1 @ vbr 256 kbps is almost always "equal" to AC3 5.1 @ cbr 384 kbps. Now i'm sure that factoring in that the original AC3s are lossy might affect the result of the mentioned tests, but how much?
Go to the top of the page
+Quote Post
saratoga
post Jun 26 2012, 02:44
Post #6





Group: Members
Posts: 4967
Joined: 2-September 02
Member No.: 3264



QUOTE (Makaveli7184 @ Jun 25 2012, 15:29) *
Well, let's say that i wanted to convert an original AC3 to another AC3 file (audio unedited), would there have been any need to go higher then the original 384 kbps of the original AC3??


Why would you do that?
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 26 2012, 03:05
Post #7





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



QUOTE (saratoga @ Jun 26 2012, 01:44) *
QUOTE (Makaveli7184 @ Jun 25 2012, 15:29) *
Well, let's say that i wanted to convert an original AC3 to another AC3 file (audio unedited), would there have been any need to go higher then the original 384 kbps of the original AC3??


Why would you do that?

  • Because i can.
  • Because i live in a democracy.
  • Because that's what the voices in my head are telling me to do.
  • Why does anyone do anything really?
  • What does it matter to you?
  • It was a theoretical question not meant to be taken verbatim.
  • Because one, for example, might need to remove 2 seconds of silence from the middle of the track (which for practical purposes translates to "audio unedited" since its effect is negligible encoding wise).
Go to the top of the page
+Quote Post
saratoga
post Jun 26 2012, 04:05
Post #8





Group: Members
Posts: 4967
Joined: 2-September 02
Member No.: 3264



If you're just doing it for fun, might as well grab a few samples and see for yourself what bitrates are needed.
Go to the top of the page
+Quote Post
probedb
post Jun 26 2012, 08:03
Post #9





Group: Members
Posts: 1232
Joined: 6-September 04
Member No.: 16817



QUOTE (Makaveli7184 @ Jun 26 2012, 03:05) *
  • Because i can.
  • Because i live in a democracy.
  • Because that's what the voices in my head are telling me to do.
  • Why does anyone do anything really?
  • What does it matter to you?
  • It was a theoretical question not meant to be taken verbatim.
  • Because one, for example, might need to remove 2 seconds of silence from the middle of the track (which for practical purposes translates to "audio unedited" since its effect is negligible encoding wise).


In the time it's taken you to sit and wait for other people to answer your very bizarre question you could have done all of this yourself.
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 26 2012, 15:55
Post #10





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



Wow... The hostility on this forum is crazy.. Second time registering here, and it seems just like the first time, i might take a long hiatus again!!
All i'm asking for is the opinion of some more experienced users regarding some matters and it's always "why do u want to do dat?" and "listen and figure it out"...
I can't for the life of me understand this collective need of withholding knowledge. I mean it goes without saying that one should do some testing of his own, but is it this hard to ask for people to share their experiences too?? That's all i'm asking for, if you happen to have an opinion on this issue, or some past experiences, drop a recommendation. No one's gonna hold you up to it.

probedb:
There is nothing bizarre about my question, and it is not a matter of time. I'm not trying to shortcut my way into taking a decision. On the contrary, i've done some extensive testing, but i'm not in a hurry, and would like to factor in some others' recommendations into my final decision.


So here, let me try this again:

Let's say i have a 5.1 AC3 audio file encoded @ 384 kbps. Now let's say i want to edit some silence in and out of this audio track, then re-encode it to the same format (5.1 AC3). What would be the "most logical" bitrate setting to use? Would it usually be enough to stay @ 384 kbps, or is there a need to go higher?? Would going higher then 384 kbps in this case be a waste of space??
Not asking for anyone to spend their time doing listening tests on my behalf. Just share what you think or if you have some past experience and that's all.

That's the whole point of lossy encoding. To achieve highest quality at lowest bitrate possible. And to rely solely on my current subjective listening experience is a flawed logic. That's why i put my trust in some more experienced users in this forum, for the second time around.

Thank you.
Go to the top of the page
+Quote Post
onkl
post Jun 26 2012, 16:14
Post #11





Group: Members
Posts: 125
Joined: 27-February 09
From: Germany
Member No.: 67444



Lossy to lossy is ugly and not so advanced formats like AC3 at very low bitrates probably suffer most. So when you re-encode your AC3 file at the same bitrate, quality will be worse. If it's still good enough for your ears, only you can tell.

Simple cutting and adding silence could be done without re-encoding.
Go to the top of the page
+Quote Post
jetpower
post Jun 26 2012, 17:30
Post #12





Group: Members
Posts: 67
Joined: 6-September 05
Member No.: 24363



QUOTE (onkl @ Jun 26 2012, 17:14) *
Simple cutting and adding silence could be done without re-encoding.

yes, but with which tool?
Go to the top of the page
+Quote Post
john33
post Jun 26 2012, 17:34
Post #13


xcLame and OggDropXPd Developer


Group: Developer
Posts: 3760
Joined: 30-September 01
From: Bracknell, UK
Member No.: 111



When decoding a lossy audio file, the content that was discarded by the encoder is replaced by noise (it may not be audible, but it's there) to pad out the data. Whatever encoder you use to re-encode that wave data will not 'know' what is real audio data and what is the added noise. To go from ac3 to ac3, you will need to increase the bitrate to reduce the quality loss, but whatever bitrate you choose, there will still be further degradation in quality. To go from one lossy format to another often appears to work better as the psychoacoustic models will differ. However, it is impossible to suggest a suitable bitrate in the example given as it will be content and audience dependent. You may get away with using a similar bitrate but it is unlikely that a bitrate much lower will be acceptable.

As has already been said, ac3 5.1 at 384kbps is already pushing the limits of what ac3 can achieve and transcoding to any other format at a lower bitrate with any level of transparency will be a challenge. Only you can determine whether the result will be acceptable to you. Most people, I would suggest, would leave the audio alone as the bitrate is not excessive by any stretch and any savings in size achieved are likely to be minimal.


--------------------
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/
Go to the top of the page
+Quote Post
john33
post Jun 26 2012, 17:40
Post #14


xcLame and OggDropXPd Developer


Group: Developer
Posts: 3760
Joined: 30-September 01
From: Bracknell, UK
Member No.: 111



QUOTE (jetpower @ Jun 26 2012, 17:30) *
QUOTE (onkl @ Jun 26 2012, 17:14) *
Simple cutting and adding silence could be done without re-encoding.

yes, but with which tool?

'delaycut'. Google it and you will find many links. wink.gif


--------------------
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 26 2012, 18:35
Post #15





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



QUOTE (onkl @ Jun 26 2012, 15:14) *
Lossy to lossy is ugly and not so advanced formats like AC3 at very low bitrates probably suffer most. So when you re-encode your AC3 file at the same bitrate, quality will be worse. If it's still good enough for your ears, only you can tell.
Simple cutting and adding silence could be done without re-encoding.


QUOTE (john33 @ Jun 26 2012, 16:34) *
When decoding a lossy audio file, the content that was discarded by the encoder is replaced by noise (it may not be audible, but it's there) to pad out the data. Whatever encoder you use to re-encode that wave data will not 'know' what is real audio data and what is the added noise. To go from ac3 to ac3, you will need to increase the bitrate to reduce the quality loss, but whatever bitrate you choose, there will still be further degradation in quality. To go from one lossy format to another often appears to work better as the psychoacoustic models will differ. However, it is impossible to suggest a suitable bitrate in the example given as it will be content and audience dependent. You may get away with using a similar bitrate but it is unlikely that a bitrate much lower will be acceptable.

As has already been said, ac3 5.1 at 384kbps is already pushing the limits of what ac3 can achieve and transcoding to any other format at a lower bitrate with any level of transparency will be a challenge. Only you can determine whether the result will be acceptable to you. Most people, I would suggest, would leave the audio alone as the bitrate is not excessive by any stretch and any savings in size achieved are likely to be minimal.

Thank you for these informative and constructive responses.
DelayCut will be of great help in some DVD authoring/backup projects.

In my current case though, audio editing might include fading in/out in addition to some cutting, so re-encoding can't be avoided. Also, i want to re-encode in AAC for HW playback reasons.
The AC3 to AC3 re-encoding question came as a benchmark comparative mean for me to choose my AAC encoding bitrate, since i now trust that AAC vbr quality preset 3 for 5.1 audio is pretty much equivalent to AC3 @ 384 kbps. However, your additional explanation of the lossy to lossy encoding flaws convinced me of taking the safe route (for me that is) and re-encoding in AAC vbr quality preset 4 (which yields on average 360~400 kbps in 5.1 audio).

This post has been edited by Makaveli7184: Jun 26 2012, 18:38
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 26 2012, 18:39
Post #16





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



Also, is there a decent "DelayCut" equivalent for MP3 and/or AAC?
Go to the top of the page
+Quote Post
onkl
post Jun 26 2012, 18:45
Post #17





Group: Members
Posts: 125
Joined: 27-February 09
From: Germany
Member No.: 67444



MP3directcut which can also do fading, so something like that might even be possible for AC3.

If size is not crucial, you can raise the bitrate to make sure quality doesn't degrade too much.
Go to the top of the page
+Quote Post
Makaveli7184
post Jun 26 2012, 19:04
Post #18





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



QUOTE (onkl @ Jun 26 2012, 17:45) *
MP3directcut which can also do fading, so something like that might even be possible for AC3.

DelayCut doesn't do fading. I'll check for similar tools. But how can a tool do fading without re-encoding?


QUOTE (onkl @ Jun 26 2012, 17:45) *
If size is not crucial, you can raise the bitrate to make sure quality doesn't degrade too much.

Well, size is important, and i was pretty happy with re-encoding to AAC vbr 3 (at around 280 kbps), so to me, going to vbr 4 at around 380 kbps is the max i'm willing to raise the bitrate to. Plus i'm encoding in VBR so i'm not too hung up on the actual bitrates (something that i learned long ago on these forums wink.gif).
Go to the top of the page
+Quote Post
Dynamic
post Jun 28 2012, 07:03
Post #19





Group: Members
Posts: 821
Joined: 17-September 06
Member No.: 35307



QUOTE (Makaveli7184 @ Jun 26 2012, 18:04) *
QUOTE (onkl @ Jun 26 2012, 17:45) *
MP3directcut which can also do fading, so something like that might even be possible for AC3.

DelayCut doesn't do fading. I'll check for similar tools. But how can a tool do fading without re-encoding?


In MP3 and AAC-LC, there's a global gain value in each chunk of audio which is in steps of approx 1.5 dB. This allows mp3gain and aacgain to modify the volume of the whole song (by walking through and raising or lowering every chunk's gain value by the same amount). mp3directCut and similar tools can boost or quieten only a selection within a song by applying the increment or decrement to only the chunks within the selection, and they can apply a "logarithmic fade" that goes from, say, 0 dB to -48 dB in steps of -1.5 dB throughout the selection. The overlapping of the chunks in the MP3 decoder smooths out any edge effects you might expect from the sudden step-wise jumps, so they tend to sound smooth without any audible popping noises.

I don't play around with AC3 much or know how AC3 is put together, but there are other codecs of a similar vintage (such as MP2) that don't feature global gain fields, so the fading might not be possible without transcoding (though you could possibly transcode only that section of audio and stitch it on to the original AC3 of the rest of it - you'd need to experiment to check for glitches at the join points)


--------------------
Dynamic the artist formerly known as DickD
Go to the top of the page
+Quote Post
Makaveli7184
post Jul 10 2012, 20:03
Post #20





Group: Members
Posts: 22
Joined: 21-June 12
Member No.: 100878



Thanks for everyone who contributed in this thread. You've been really helpful. I now have a better understanding on transcoding in general, and avoiding it when possible thanks to tools like delaycut and mp3directcut.

Thanks again.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 19th September 2014 - 23:04