[Honus]

I am presently ripping everything MP3 @ 320kbps (very easy). Would ogg vorbis provide better SQ, and what concesions will I make switching to ogg vorbis? My rig is iaudio G3 and UE superfi 5pro.


Thanks!

 

[wanderman]

vorbis lancer rips faster then lame and reaches transparency at a much lower bitrate then lame, but vorbis is more tasking on the battery and dosen't have as much widespread compatabily as mp3. But vorbis is the better codec then lame when only comparing size/performance.

 

[Brent Hutto]

Quote:

Originally Posted by Honus I am presently ripping everything MP3 @ 320kbps (very easy). Would ogg vorbis provide better SQ, and what concesions will I make switching to ogg vorbis? My rig is iaudio G3 and UE superfi 5pro.

There's no evidence that Ogg Vorbis at 320kbps VBR is distinguishable from MP3, AAC or WMA at the same bitrate. Or for that matter distinguishable from FLAC or WAV. If you have well-encoded files (such as the ones you get from the latest LAME with proper settings) in 320kbps MP3 then play the heck out of them and enjoy your music. If you don't like how it sounds, the problem ain't your choice of encoding format.

[EDIT]Given that the available evidence, such as it is, doesn't really go very far toward addressing this issue perhaps I should have said "In my opinion there's no reason to believe Ogg Vorbis at 320kbps VBR is distinguishable..." rather than making the original flat assertion to that effect.

 

[blessingx]

There's no evidence it isn't. The repeated 320 kbps LAME (or even lower bitrate) is indistinguishable from uncompressed often sites HydrogenAudio related tests which never state equipment used. It's their inexcusable large blind spot. So if a dozen people test a dozen tracks on unknown equipment I'm not sure I'd use those results to dictate the limits of human hearing (yours or others). This gets repeated often and I don't care to rehash a larger fight.

There's no reason to think 320 kbps (or even 500 kbps) Ogg Vorbis wouldn't sound better than the highest LAME, but it's unlikely that iAudio DAC is the best around, nor the environmental conditions you're likely to use it be silent, so I certainly wouldn't waste the space - or possibly even more important - battery life sacrifice for it. Then again I think the 320 kbps is overkill (and only selected because "it's the highest") for portable uses. If you're going to compromise for space (lossy), I say compromise for space (~192-224 VBR kbps AAC or Vorbis).

 

[pata2001]

I would stick with MP3 for the sake of compatibility. There's no use of using a format that is only usable with a few players, regardless of quality. With MP3, I wouldn't have to worry about future DAP upgrades.

 

[D555]

I think volumes have been written on the subject.

I'm certainly impressed with Ogg Vorbis (aoTuv r1) and am amazed at the sound that comes from the lower bit rates: -q4, -q5, -q6

I like LAME as well (trying 3.97b3) but I think I hear a few more artifacts for a given bit rate as compared to Ogg.

Just my view, but my listening of the current speed/size options (VBR-new, joint stereo) for LAME are at the the expense of sound.

 

[Honus]

Thanks for your thoughts. I am very new to this. I was exclusively a CD user up to a few months ago. I've been using Windows Media player to rip all my tunes. Is this approach greatly different form LAME encoding or are they one in the same? One thing I would hate to give up is the automated tagging Windows provides. However, if ogg were noticebly better in terms of SQ, I would consider making concessions. Your responses to my query would indicate that I should stay the course.

 

[kwohy233]

Windows Media Player rips in wma format. Check your settings- its probably on 64kbps or 128kbps. In terms of sound quality? I would say instead of asking others, you should get some cds you know well and rip at various formats and see for yourself at what level you stop being able to tell the difference between bitrates, and compare the variou formats.

Personally, I go with LAME mp3 V2 or V3 because Ive found ogg significantly reduces battery life, and on my samsung yp-u1 player, there seems to be a huge increase in the background hiss. And who knows? oneday you might buy an ipod... which means if you chose ogg you'll have to re-rip.

 

[Febs]

Quote:

Originally Posted by blessingx There's no evidence it isn't. The repeated 320 kbps LAME (or even lower bitrate) is indistinguishable from uncompressed often sites HydrogenAudio related tests which never state equipment used. It's their inexcusable large blind spot.

That information is often available. For example: http://ff123.net/setup.html

 

[Klefbek]

Quote:

There's no evidence it isn't.

So therefor it is? If everything that isn't proven is automatically to be assumed true, you are opening many cans of worms. I am not aware of any evidence that Vorbis at ~320 kbps is better than LAME.

Quote:

The repeated 320 kbps LAME (or even lower bitrate) is indistinguishable from uncompressed often sites HydrogenAudio related tests which never state equipment used. It's their inexcusable large blind spot. So if a dozen people test a dozen tracks on unknown equipment I'd use those results to dictate the limits of human hearing (yours or others).

a. It is not always true (thanks Febs)
b. Even when it is given I am sure your reply will be that the test was not relevant to you because you have different (and no doubtedly, superior) equipment.
c. No test at HA.org is claiming that 320 kbps LAME or any other codec at any bitrate is transparent in all cases, for everyone, on any equipment. In fact, there are threads there that quite show the contrary. Perhaps the problem is the people reading the tests, not the tests themselves.

@Honus:

In your case I doubt you would hear any improvement in SQ by switching to Vorbis

 

[blessingx]

Quote:

Originally Posted by Febs That information is often available. For example: http://ff123.net/setup.html

That's one individual. I could be wrong (as I hadn't seen this) but I've never seen the list of equipment from the test group. If this is available I'd love to see it for once. Could someone please link to it? These old tests still get used often and I don't see a listing there. Anyone got a link?

Quote:

Originally Posted by Klefbek So therefor it is? If everything that isn't proven is automatically to be assumed true, you are opening many cans of worms. I am not aware of any evidence that Vorbis at ~320 kbps is better than LAME. a. It is not always true (thanks Febs) b. Even when it is given I am sure your reply will be that the test was not relevant to you because you have different (and no doubtedly, superior) equipment. c. No test at HA.org is claiming that 320 kbps LAME or any other codec at any bitrate is transparent in all cases, for everyone, on any equipment. In fact, there are threads there that quite show the contrary. Perhaps the problem is the people reading the tests, not the tests themselves.

Hydrogens tests are quite interesting and I initially use to visit there as much as here. I think they're quite valuable, but yes my problem is that readers take it as gospel (limits of human hearing) primarily (I suspect) because the sites posting rules history. There's a general net groupthink (at HA, here, Slashdot, etc.) that because of ABX testing their results are scientifically sound (even with the above important hole **) and can be applied far outsite their testing scope (for example that a 128 kbps test apples to 224, etc.). I'm not going to go into this deeply as previously because if you use the search tool you'll see this has been argued to death already over and over.

Please read the order of my posting. I think there's a danger in posting (and note the reference of others repeating) that 320 LAME (or any bitrate as many testers have found 128 kbps to be transparent on many tracks) is indistinguishable from uncompressed without proof (and your 'c' seems to indicate my comment is valid) for someone else (who may have better hearing) without reasonably conclusive data across the board.

We can make assumptions in this case because (as I mentioned) the iAudios DAC and likely listening environment won't be that most demanding.

** And yes the tests are always framed as "encoder" or "codec" tests. When are they clarified within a subset (age range, gender equality [which is nearly as important as age with audio limits], etc.)? This is only important in how the results are perceived and Hydrogen is SO concerned with perception (even shutting down the site for while once) that I'm surprised at least the equipment isn't addressed in tests posted (even if the tests are only loosely associated with HA, HA is pretty stringent in post rules). Again they're some of the most sound tests out there ... just not are conclusive as their siters keep implying.

 

[Klefbek]

Quote:

** And yes the tests are always framed as "encoder" or "codec" tests. When are they clarified within a subset (age range, gender equality [which is nearly as important as age with audio limits], etc.)? This is only important in how the results are perceived and Hydrogen is SO concerned with perception (even shutting down the site for while once) that I'm surprised at least the equipment isn't addressed in tests posted (even if the tests are only loosely associated with HA, HA is pretty stringent in post rules). Again they're some of the most sound tests out there ... just not are conclusive as their siters keep implying.

It is not so difficult why equipment is not in TOS#8. It really does not matter on the larger scheme of things. An ABX test can only prove if an audible difference exists, not that it does not exist (the difficulty with proving a negative). In other words, in individual listening tests negative results are quite unintresting because nothing has been proven, so other exact conditions (like equipment and age) do not matter. In case of a positive result, the exact conditions do not matter either, because we have only the codec to blame for the audible difference. Indeed, because a condition of a proper ABX test is that both source and encoded file should be played on the same equipment. IOW: the factor equipment, nor age, nor anything else you can think of does not add or change anything about the presented result.

For a larger scale (not individual) listening tests factors like equipment age and gender are only useful to know if they are controlable variables. If every participant would have given these details they could not in anyway be used to make the tests more accurate.

When is knowing equipment useful? When the same samples are used, with the same people on equipment setup X1, X2 ... Xn.
When is knowing age useful? When the same samples are used, with the same equipment, on different 'sets' of people, each with a different age range.

HA listening tests are not organised in this way. Perhaps they should, but at best I think it remains very unclear how these factors would really influence the result, if at all. I can ABX LAME 320 kbps on some samples with $5 headphones. Not that those headphones are great, or LAME is pretty bad, it was just that those samples prove to be very difficult.

Personally, I believe that sample selection, listener training, and listening environment are factors that weigh much more than equipment or age.

 

[Brent Hutto]

So if I get 100 people to do a set of ABX tests with their iPod earbuds then that's as valid as if the same 100 people had done the tests with US$10,000 audiofile setups and US$500 headphones?

In each of these example scenarios all 100 people are using the same equipment so there's no need to "control for" the fact they're using earbuds. However, results of the iPod earbud tests aren't going to be very interesting to people with decent headphones. That's a validity problem, not a precision or accuracy problem.

 

[K2Grey]

Like Klefbek said, if someone's test is negative it doesn't mean anything. There are always people with better setups or who claim to have better ears.

If someone's test is positive, then it doesn't matter what equipment is used. The fact their test comes out positive means their equipment is good enough to hear the difference with.

That is why in both cases, it is not meaningful to say what equipment was used.

Now, if you think a negative test does mean something, then it would be important what equipment was used. But again, it does not mean anything, since even if 30 HydrogenAudio members did ABX tests between LAME v0 and FLAC with $10,000 systems, you would still see people on the forums claiming that HA members all have tin ears and that FLAC is 50x TIMES BETTER OMG!!!!! than LAME mp3 when played on iPod -> UM2. And the only way to shut him up is to get him to ABX it himself and see whether he gets a positive result or not.

 

[Brent Hutto]

So if I understand correctly, the only valid information issuing from the HA forum is as follows: "You ought to ABX".

Any actual information about how encoders or bitrates compare to each other is by definition entirely specific to the particular equipment, listener, conditions, phase of the moon and feng shui of each individual test. Therefore it is not useful to mention any of those particular factors since even if you knew all of that the test wouldn't be reproducible.

In that case, there's no difference at all between valid and invalid results. You could replace the whole forum with a set of instructions on how to conduct an ABX test and then let everyone conduct them in their own bubble.