IPB

Welcome Guest ( Log In | Register )

2 Pages V  < 1 2  
Closed TopicStart new topic
From FLAC to v2 VBR - 2 Issues, Basic encoding advice
bernhold
post Apr 27 2013, 01:08
Post #26





Group: Members
Posts: 41
Joined: 22-March 13
Member No.: 107330



You want me to deliver proof of what I said, and when I do so, you simply refuse to have a look at that proof? Or dismiss it as "crude" without any basic effort of disproving it? What makes a 'crude' (in neutral terms, simple) example worthless in your eyes? Anyway, regarding to the original question, our discussion is going off-topic. It's interesting and I value your opinion but we should continue to discuss it in a separate thread sometime.
Go to the top of the page
+Quote Post
greynol
post Apr 27 2013, 01:22
Post #27





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



I thought I already accepted that you found a pair of headphones which you deem to be of "higher quality" that makes it easier for you to hear artifacts for the samples you've attempted to ABX.

http://www.hydrogenaudio.org/forums/index....st&p=832045

...rinse and hopefully not repeat.

This post has been edited by greynol: Apr 27 2013, 01:30


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
Gvaz
post May 31 2013, 10:41
Post #28





Group: Members
Posts: 21
Joined: 11-November 06
Member No.: 37425



I do the same thing as you op, but I do V0

Here is the code I use on mine:

-S --noreplaygain -V0 -ms - %d
Go to the top of the page
+Quote Post
db1989
post May 31 2013, 10:55
Post #29





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



QUOTE (Gvaz @ May 31 2013, 10:41) *
-ms
And once again, I must advise against using simple stereo (-ms), for reasons that have been discussed to death in previous threads. Suffice it to say that joint stereo is default for a reason, namely its mathematical losslessness and promotion of efficient compression, and that any supposed objections to it are based upon misconceptions and FUD. If you are going to recommend things, be sure your reasoning is valid.
Go to the top of the page
+Quote Post
Gvaz
post May 31 2013, 11:48
Post #30





Group: Members
Posts: 21
Joined: 11-November 06
Member No.: 37425



From what I can tell, joint is an interpolation of data whereas having it as simple stereo keeps everything "as is". Which is more to what I'm looking for, an unchanged lossy version of the data being converted into a smaller footnote.
Go to the top of the page
+Quote Post
db1989
post May 31 2013, 11:54
Post #31





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



QUOTE (Gvaz @ May 31 2013, 11:48) *
From what I can tell, joint is an interpolation of data whereas having it as simple stereo keeps everything "as is".
What you can tell from where? In any case, your interpretation is incorrect. Mid/side encoding is mathematically lossless. This is what LAME uses for joint stereo, and it automatically chooses between simple and mid/side modes on a per-frame basis so that each frame is encoded most efficiently. Thus, by forcing simple stereo, you are crippling the ability of the encoder to operate with maximal efficiency. All in the name of concepts you do not understand.

QUOTE
Which is more to what I'm looking for, an unchanged lossy version of the data being converted into a smaller footnote.
What does this even mean? “unchanged lossy”? How is that not a contradiction in terms? Once again, mid/side encoding is mathematically lossless, and the fact that it can often save space when compared to simple stereo means that the subsequent packing of each frame into however many bits the encoder allocates it can be done more efficiently because bits are not being wasted on an inefficient method of compressing stereo data.

Again, please do try to know what you are saying before you say it.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 12:48
Post #32





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



@bernhold: the point you're missing is what real life encoding artifacts sound like. Equipment really doesn't matter. It can happen that on a low value equipment artifacts are easier to hear. Sure for a specific issue different equipment has an influence on audibility, but there's no tendency for 'high value equipment makes it more audible'.

What matters much more is quiet environment or not and consciously listening to the music or not. I guess most people listen most of the time to music not as their primary activity, and in such a situation encoding quality needn't be at maximum, especially as even -V5 usually yieds a very good quality (see 128 kbps mp3 listening test).

I personally dislike suggestions like 'do some ABXing and try for yourself'. While this isn't a bad advice my feeling is that this isn't reaaly helpful. On one hand many people don't like to do the pain of ABXing. And ABXing as the decision criterion is very questionable. It's easy to provide samples which show that say -V2 can provide bad quality. But what this tells is always about exactly these samples. On the other hand an attitude like 'don't care about special samples which are encoded badly' may not calm things down for everybody. We are all different, and everybody should feel fine personally which comes from personal attitude where any argument fails.

I think most helpful for a decision is
a) the beforementioned listening test as well as a personal listening test at 224 kbps. As for the latter be aware of that it contains critical material, and the tester is supposed to have an exceptionally good hearing and did put a lot of effort into this test.
b) thinking about how hard storage restrictions are. Why not use the highest encoding quality you can allow for with respect to storage space restrictions?

This post has been edited by halb27: May 31 2013, 12:49


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
mjb2006
post May 31 2013, 14:00
Post #33





Group: Members
Posts: 787
Joined: 12-May 06
From: Colorado, USA
Member No.: 30694



Gvaz, any stereo signal can be expressed as left & right, or mid & side; it doesn't matter, one is not any more or less accurate than the other. The advantage of mid/side is that the side channel is often quieter and simpler, much easier to compress when the left and right are fairly similar.

Regardless of whether simple or mid-side is being used for a given frame, an encoder must make a decision about how much room to allocate to the data for each channel. Ideally it will predict or test what combination will yield the best results, space-wise, and/or quality-wise if the coding is lossy. Joint means "switches between simple and mid-side at the encoder's discretion." MP3 encoders have been making very smart decisions about these issues for about a decade, now.

This post has been edited by mjb2006: May 31 2013, 14:02
Go to the top of the page
+Quote Post
greynol
post May 31 2013, 14:36
Post #34





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Personal listening tests are the ONLY way to determine which bitrate gives acceptable sound quality results for any particular individual.

Attempts to give alternate methods equal footing are futile. db1989 hit it on the head earlier.

While halb27 feels justified in proselytizing his approach, he is simply wrong. He is also wrong for trying to insert his beliefs in any thread where he sees an opportunity.

http://www.hydrogenaudio.org/forums/index....st&p=832476
Refer to the previous post for the context of this refutation that was basically (conveniently?) ignored last time around.

This post has been edited by greynol: May 31 2013, 19:46


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 21:33
Post #35





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



QUOTE (greynol @ May 31 2013, 15:36) *
Personal listening tests are the ONLY way to determine which bitrate gives acceptable sound quality results for any particular individual.

Do you mind describing a listening test procedure to make the OP feel fine with his a priori bias towards -V2, or find another quality level which better fulfills his needs?
How many samples tested will be enough to make him feel good? What genres? Only those he usually listens to? Or also genres he seldom listens to? Should he take into account well-known critical kind of music? Shall he totally ignore it and wait until he comes across such an issue with his personal collection? (Sure re-encoding is always an option, but is it the preferred option to everybody?)

This post has been edited by halb27: May 31 2013, 21:48


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
greynol
post May 31 2013, 21:35
Post #36





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



It's in the discussion I linked, even the stuff you've added in subsequent edits. It's all there, halb27; it needn't be here too, in case I wasn't clear enough about it earlier.

QUOTE (halb27 @ May 31 2013, 13:33) *
the right choice for him

A double blind test is the only fit. As to catering towards a priori biases, I think you've forgotten that this is HA.

This post has been edited by greynol: May 31 2013, 21:52


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 21:51
Post #37





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



A double-blind test for sure. But what precisely should he test? Please mind the edits in my post (sorry they were late).

This post has been edited by halb27: May 31 2013, 21:52


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
Kohlrabi
post May 31 2013, 21:52
Post #38





Group: Super Moderator
Posts: 1024
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (halb27 @ May 31 2013, 13:48) *
I personally dislike suggestions like 'do some ABXing and try for yourself'. While this isn't a bad advice my feeling is that this isn't reaaly helpful. On one hand many people don't like to do the pain of ABXing. And ABXing as the decision criterion is very questionable. It's easy to provide samples which show that say -V2 can provide bad quality. But what this tells is always about exactly these samples. On the other hand an attitude like 'don't care about special samples which are encoded badly' may not calm things down for everybody. We are all different, and everybody should feel fine personally which comes from personal attitude where any argument fails.
If one really cares so much about artifacts in some rare cases, why not go lossless? With MP3 you never have any security that all the artifacts are imperceivable.

QUOTE (halb27 @ May 31 2013, 13:48) *
a) the beforementioned listening test as well as a personal listening test at 224 kbps. As for the latter be aware of that it contains critical material, and the tester is supposed to have an exceptionally good hearing and did put a lot of effort into this test.
That test resulted in a well-expected 4-way tie. IMHO nothing was gained from that test, other than proving again that those encoders are perfectly making use of what MP3 can offer. Admittedly though, it is good that it was shown with (good) statistics once. Singular rare problem files are rightfully just outliers in the statistics, and don't change the conclusion that all encoders are equally very good. And again, if one cares so much about these rare cases, why not use lossless encoding or another lossy codec for these files. Nobody is forcing you to only use MP3.

QUOTE (halb27 @ May 31 2013, 13:48) *
b) thinking about how hard storage restrictions are. Why not use the highest encoding quality you can allow for with respect to storage space restrictions?
That's a weird way to use lossy encoders. What do you do when you expand your collection? Reencode everything to the next lower setting, just because you arbitrarily chose one based on whatever disk space you had previously? And again, instead of using lossy encoders well beyond transparency, I'd strongly recommend to just use lossless. Though I am aware of terrible hardware which only plays MP3, but I guess those devices are only used in louder environments anyway (i.e. cars).

This post has been edited by Kohlrabi: May 31 2013, 21:59


--------------------
Ceterum censeo Masterdiskem esse delendam.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 22:03
Post #39





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



QUOTE (Kohlrabi @ May 31 2013, 22:52) *
QUOTE (halb27 @ May 31 2013, 13:48) *
a) the beforementioned listening test as well as a personal listening test at 224 kbps. As for the latter be aware of that it contains critical material, and the tester is supposed to have an exceptionally good hearing and did put a lot of effort into this test.
That test resulted in a well-expected 4-way tie. ...

The point in this context (which isn't about encoder choice) is: a 224 kbps setting is very good even for critical stuff - and already 128 kbps yields a good quality.
Lossless encoding wasn't the OP's question. He came from there and knows about lossless.

This post has been edited by halb27: May 31 2013, 22:04


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
Kohlrabi
post May 31 2013, 22:04
Post #40





Group: Super Moderator
Posts: 1024
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (halb27 @ May 31 2013, 23:03) *
QUOTE (Kohlrabi @ May 31 2013, 22:52) *
QUOTE (halb27 @ May 31 2013, 13:48) *
a) the beforementioned listening test as well as a personal listening test at 224 kbps. As for the latter be aware of that it contains critical material, and the tester is supposed to have an exceptionally good hearing and did put a lot of effort into this test.
That test resulted in a well-expected 4-way tie. ...

The point in this context (which isn't about encoder choice) is: a 224 kbps setting is very good even for critical stuff - and already 128 kbps yields a good quality.
With that I can agree. smile.gif

QUOTE (halb27 @ May 31 2013, 23:03) *
Lossless encoding wasn't the OP's question. He came from there and knows about lossless.
That doesn't change the fact that if one has concerns about the lossy encode or no size restrictions to stick to lossless. Nice try to pretend I committed a ToS #5 violation, though. laugh.gif

This post has been edited by Kohlrabi: May 31 2013, 22:15


--------------------
Ceterum censeo Masterdiskem esse delendam.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 22:15
Post #41





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



QUOTE (greynol @ May 31 2013, 22:35) *
It's in the discussion I linked, even the stuff you've added in subsequent edits. It's all there, halb27; it needn't be here too, in case I wasn't clear enough about it earlier.

Sorry, I can't find answers to this:

How many samples tested will be enough to make him feel good? What genres? Only those he usually listens to? Or also genres he seldom listens to? Should he take into account well-known critical kind of music? Shall he totally ignore it and wait until he comes across such an issue with his personal collection? (Sure re-encoding is always an option, but is it the preferred option to everybody?)


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
greynol
post May 31 2013, 22:21
Post #42





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



No need to be coy. Again, it's there (and elsewhere); and yes it is an iterative process. This will be the last off-topic exchange on the matter.

This post has been edited by greynol: May 31 2013, 22:24


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
halb27
post May 31 2013, 22:29
Post #43





Group: Members
Posts: 2429
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



QUOTE (Kohlrabi @ May 31 2013, 23:04) *
That doesn't change the fact that if one has concerns about the lossy encode or no size restrictions to stick to lossless. ...

We don't know why the OP encodes to mp3. There's more to lossy encoding than just size restrictions. I personally could use lossless (or my preferred lossy codec lossyFLAC) on any of my mobile devices if it were only about storage space. But my mobile phone wouldn't play it, and my wife's iPod not either.
As for storage space considerations sure you have to try to find out what your needs are in the future. But this has also to do how to use your lossy collection (ever growing, periodically selective, or what?).
These are considerations to take into account, as well as the fact that depending on the devices used it may not be a problem to increase storage space.
To me personally storage space isn't an issue at all, and if it's not for me, I'm sure there will be other people too who needn't care. Not everybody of course. Also we don't have to worry much about storage space needs in say 10 years. Be it good or bad, but we will use other devices then with even lower storage space restrictions than we have now.


--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
Gvaz
post Jun 1 2013, 02:10
Post #44





Group: Members
Posts: 21
Joined: 11-November 06
Member No.: 37425



I can't speak for the OP, but for me FLAC takes up too much space. The audio quality between V0 and flac for me is negligible in my ability to tell the difference. Ultimately, I want a smaller filesize of an audio file while still retaining the max quality possible.

This post has been edited by Gvaz: Jun 1 2013, 02:11
Go to the top of the page
+Quote Post
db1989
post Jun 1 2013, 14:32
Post #45





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



“Max quality” is a moot point beyond the particular setting that would produce perceptual transparency for a given input stream. Something is either transparent or not. Not hearing a difference between a high setting like -V0 and lossless is the norm for most listeners. The two are then equivalent, not just negligibly different.

More generally, I’d like to split this thread on account of the several highly divergent discussions that have emerged, but I don’t know where to begin. Can we stop messing it up with even more off-topic posts?
Go to the top of the page
+Quote Post
Kohlrabi
post Jun 1 2013, 15:56
Post #46





Group: Super Moderator
Posts: 1024
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



It seems that the OPs question has been answered, and that there is no need to further discuss the well known advantages or disadvantages of lossless vs. lossy encoding or encoder choices, since the MP3 forum also has a sticky post about suitable encoder settings. I also agree with db1989 that it's very hard to split this topic into logical chunks, so to avoid further derailings, this topic will now close. If any party objects, especially IDefyAxioms, please send me a PM, and I will consider reopening it.

This post has been edited by Kohlrabi: Jun 1 2013, 15:57


--------------------
Ceterum censeo Masterdiskem esse delendam.
Go to the top of the page
+Quote Post

2 Pages V  < 1 2
Closed 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: 30th August 2014 - 12:28