I'd like to know if the quality improvement that V0 provides is worth the extra space.

Does V0 actually produce better sounding files than V2, or is V2 already transparent? Is V0 just for people who need to see a higher bitrate on the screen or is it a worthwhile preset which should increase the quality of files encoded with it compared to V2?

I've heard a lot of conflicting opinions on this one, so I'd like to hear the opinions of the MP3 experts here.

Thanks.

How does it sound to you? Can you tell the difference?

Generally speaking its overkill. Actually I don't know if anyone has found a sample that is transparent at -V0 but NOT at -V2. In theory I guess there could be, but in practice -V2 is transparent on ~99% of stuff, and that other 1% isn't transparent at any bitrate.

Of course, everyone hears things differently. Try an ABX test! You'll probably find even -V5 transparent most of the time.

See this nice graph from the HA wiki:


http://wiki.hydrogenaudio.org/index.php?ti...ame-chart-2.png

From V3 upwards, quality benefits become smaller and smaller. V2 should be a good safety margin, while you can probably use V3 or even V4 for a lot of samples.

I'm no "MP3 expert" but to my ear -V2 is pretty transparent already. I agree with Jebus, -V0 is overkill.

How we think it sounds means nothing. How it sounds to you is what matters. Ask us our favorite ice cream in an effort for you to choose one. It is about the same thing. Listen to both and see. You may find, as have most, that V2 is fine. And then again, maybe not.

Here's my take on it:

If you are using MP3(or any lossy compression), you care about file size to some degree, else you would just FLAC/WavPack everything.

Album X may be 480 MB stored losslessly.

At CBR 320, it's 200 MB.

At -V 0, its 130 MB.

At -V 2, its 100 MB.

CBR 320 seems bad to me, simply because it has no clear advantage. It's very large, to the point where the whole point of lossy is downplayed. Quality may not really be much if any better than V2/V0, even though it theoretically cannot be worse.

Size-wise, V0 is where I find my home for MP3. You can say, "But V2 would likely have been transparent as well." But it matters not, because once the compression is down to what -V 0 supplies, I have already satisfied my need for "make things smaller". And since V0 cannot(in theory) sound worse than V2, and has the pottential to sound better, I stick with it.

As for you, if you care at all about the size difference between V2 & V0, you should definately use V2. If the difference in size between them is not a concern to you, use V0. I would not recommend 320CBR, since at that point, you might as well use lossless.

And, of course, if you are using these MP3's in a car or any other noisy environment or portable player, you can be guaranteed pretty much that even V2 is way overkill, let alone V0. For portable use, I use V5. Seeing what other people do is great to use as a benchmark to see where you might want to go, but ultimately you must listen and decide for yourself, else it's pointless.

[Edit: Actual file sizes will vary, but you can use the numbers I chose as rough guidelines.]

In my opinion -V2 is a good safety margin (as already explained) as it has a target bitrate of 190 kbps which is why I use it, that and the old --preset standard recommendations from not long ago still have significant influence upon me hence the reason I use -V2 --vbr-new on everything. Although I must state that to my ears -V4 and -V3 are transparent to me.

I agree with you to the most part, but seeing as how you're saving 30Mb's with V2 and it's transparent for 99% of everything, I stick with that...simply due to the fact I have a ridiculous amount of music on my HardDrives. 30Mb's for each album can mean I can fit quite a bit more on them.

It may just be me but I find that when I encode in V0 (Lame 3.97) it sounds louder than V2 at the same volume

Assuming you're using the same program/method to encode it, that's almost certainly placebo. ABX it to make sure.

I don't mean to criticise, but that's not a very efficiently constructed/scientific graph. I'll have a go at producing an alternative one when I get back home.

What do you dislike about this graph?

I do hope the alternative graph makes better use of known scientific methods (i.e. TOS #8) in audio testing, if the current one is 'unscientific'.

By all means have a shot at building a new graph!

This one, I think, can only be thought of as *indicative*. With such a thing as quality, the differences between each person's ears can be like night and day. Much of this percieved high-bitrate 'quality' comes from the ancient 'fact' that stuff around 128k is supposed to sound terrible and the thought "I'm an audiophool, I need much higher bitrate than those primitives" (subconciously, of course). How refreshing (demotivating for some) an ABX test can be here.

The best thing to build another graph like this would be to better document how the 'quality' scores were evaluated (especially beyond V2, where practically no-one, with non-killer-samples, can ABX - 128k seemed mostly transparent already in recent listening tests).

A well encoded 192kbps file is already much transparent to me. All our FUD is placebo.

With 3.98 alphas there is an improvement when going from -V2 to -V1 IMO for critical parts of the music.
-V0 doesn't bring an additional substantial improvement however.

Several years ago I encoded my entire CD collection in what was then the equivalent of -V0. It took up 25 GBytes of my 40 GByte hard drive, so you could probably say that I should have used -V0 instead since I would almost certainly never hear the difference. Since then I have upgraded computers and gotten into video, and my present system has more than 1 TByte of disc space. My 25 GByte music collection is now a very minor fraction of my storage space, making my original choice look much better.

So the bottom line to me is that it probably doesn't make a lot of difference these days whether you choose -V0 or -V2 or even lossless, the storage space isn't such a big deal and people have more important things to worry about. So just go with whatever makes you happy. At least that's my opinion.

True with computer storage, not so true with portables, especially low capacity flash models.

I use V0 because I play the same files on my PC and my iPods - and on my home system V1/V0 makes a difference. But if I used V2/V3 on the portables I could store whole lot more songs with little or no perceived loss in quality.

you mean you can abx it, or that you "think" it does?

I have 120 GB of MP3s now, at -V4. I wonder how much that would be at -V0 ...

I mean that I don't want to get into semantic sniping over ABX.

Soon portables will be like 100GB. Soon... but not right now. Then it will be FLAC, or at least WAV. (Most already support WAV).

Well I have a lot of samples what I can feel the diference with -V2 and not with -V0.
Songnames:Angra - Spread your fire [begining when the drums come at the the music, there is a lot of pre-echo with V2]
Savatage - Visions [ in the this track i can ABX'ing easily all this at V2 and hardly\not can't tell the diference at V0]
Queen - I'm Going Slightly Mad [an strange sound at about 20 seconds]
Iron Maiden - Purgatory [the guitar's sound at begining before the first beat of drums, the diference is more annoying in the right channel, at V0 all the diferences gone]
and more.....

Im quite happy around v5!

V5 have excellent quality compared to the compression ratio aprox 11'30 in one cd, but what is questionated is the gain of quality going of V2 to V0.

If I were going to raise the quality from V2, I would use ABR 240/250 or something instead of using V0 (see halb27's threads and one about LAME 3.98a10/11). ABR appears to have better results than VBR at high bitrates.

One further thought on choosing amount of compression for portables.

Battery life depends largely on amount of hard disc activity. The hard drive activity depends on how fast you drain the audio cache. So larger files will drain your battery faster. (Thirty minutes of WAV files is a lot more disc action than thirty minutes of -V2 mp3 files.) I'm guessing that battery hours might be inversely proportional to file size. However, I am no expert on this, and I have not even run experiments with it yet.

Perhaps others have some direct experience with this, and would like to share it.

I just received an 80G iPod as a gift, and I could easily fit my entire collection on it as WAVs. But I fear this would run the daylights out of the hard disc. Encoding at -V0 really squashes size down. Going to -V2 is a little bit better, but not really substantial. So, from a battery-life standpoint, I could go with either -V0 or -V2, with probably little difference.

Any thoughts?

How about Apple Lossless as a compromise - all the quality of WAV, 1/2 the size.

One really big iPod battery drain on the hard disk models is shuffle - jumping around on the disk and flushing/refilling the look ahead cache. Playing straight through contiguous sectors (an album) seems to use far less battery.

I'm not sure this is true. MP3 and other compressed formats (both lossy & lossless) must be decoded to PCM for playback, whereas WAV is simply a container for PCM and thus requires no decoding. I might be wrong here, but I think that the resources used for decoding would have a greater effect on battery life than hard drive activity.

I do agree (and recall reading here at HA) that, when comparing compressed files of the same format, higher bitrates will drain battery life faster than lower ones.

Lastly, I'd just like to voice my opinion that using uncompressed WAV, lossless, or even LAME -V0 on a portable is a plain silly waste of space.

The resources required by spinning the drive is a lot higher than the resources required by decoding. That's why iPods have 32/64MB buffers, and why they say that if you skip tracks a lot, you'll greatly reduce the effective battery life.

Pleeeeeeeeeeease, upload that clips!!!

Assuming your album tracks are on contiguous sectors.

What happens when you take albums off and put different albums on, do all the tracks stay contiguous?

The iPod might be smart and might preload related tracks into the cache regardless of location on the HDD. (I don't actually know, I'm venturing a guess)

This is almost off topic.. I did som ripping last year and I tried to find the best setting qual/size V0 and V2 compared and I used the -verbose var. and saw the following:


And when you read the docs it states that shape 2 is well enough for human ears and that shape 1 was overkill.. But still.. I chose V0.. : )

Edit: I cant seem to get the tabs work properply..

The iPod gets fragmented - like most disks and it takes more work to get the next track. As someone else pointed out, skipping tracks is also very costly because it keeps the disk spinning.