My Experience with WMA Lossless and the Implications Thereof

[Damion Chaplin]

For quite some time now I’ve been wondering whether I shouldn’t be ripping my CDs in a lossless format in case something happens to the original CD. I’ve always used Exact Audio Copy with LAME MP3 to produce my MP3s and have always been happy with the results. I knew that EAC was capable of ripping my CDs into lossless formats, so this weekend I decided to bite the bullet and give Microsoft’s lossless WMA format a try. My results both dismayed and delighted me, and I thought I would share them here, since I knew at least a few of you would be interested.<br /><br /><img src="http://www.digitalmediathoughts.com/images/Hand_CD.jpg" /> <br /><!><br /><span><b>Lossless? Why?</b></span><br />First, a little history: I don’t consider myself an ‘audiophile’--more of a ‘music lover’ really--and frankly I’ve always thought a high-bitrate MP3 was ‘good enough’. My wife is a big fan of Annalee Newitz, a proponent of lossless digital audio, and has often asked me “But don’t MP3s throw away data?” I then explain to her that the real-world difference between a 256 kBit/s MP3 and a CD were too small for either of us to hear. My wife’s not that technologically-inclined, so I didn’t bother explaining that lossless formats also take up a lot of space and don’t work with the majority of portable media players. Or that it would take a really long time to re-rip all my CDs, turning my 60GB collection into who knows how large? So for more than a year now, I’ve been resisting the idea of ripping to lossless.<br /><br />All the while though, there was the voice in my hindbrain telling me “But there must be an audible difference between a CD and MP3, no matter how high the bitrate. That’s why they call it ‘compressed’.” Well, this weekend I found myself with that rare thing called ‘free time’ and decided it was time to give in to the voice and at least give it a shot. I’m very glad I did.<br /><br /><img src="http://www.digitalmediathoughts.com/images/ExactAudioCopy.jpg" /> <br /><br />As I’ve stated, I’ve always been a huge fan of <a href="http://www.exactaudiocopy.de/">Exact Audio Copy</a>, a free CD ripping program that does exactly what its name says and no more. Together with the also-free <a href="http://mitiok.cjb.net/">LAME MP3</a> compression codec, it can be a very powerful tool for those wishing to get the best-sounding MP3 files from their music CDs. Although it’s been possible for some time now, I’d only recently learned that EAC is capable of ripping CDs into lossless formats too. I figured if EAC could do it, I should at least see what happens.<br /><br /><b><span>The Decision</span></b><br />The first decision I had to make was to decide on a lossless format. I should probably mention here that I don’t own an iPod, don’t use iTunes and really don’t plan to in the foreseeable future. That narrowed a very short list down to the two main competitors: FLAC and WMA Lossless. I had known that this tough decision would be ahead of me, but that didn’t make it any easier to choose. I had experience with FLAC and had never used WMA Lossless before. FLAC is supported on some players, but no current PMP supports WMA lossless (except the ill-fated Portable Media Center). In the end I decided on WMA Lossless, a decision mostly based on instinct. I guess I just have more faith that Microsoft’s format will be around longer. In addition, my current PMP, while not supporting WMA Lossless, does support WMA files. I can still hold out hope that they release a firmware update that gives my player lossless capability. It’s a slim chance, but much better than them giving it FLAC support! :) <br /><br /><span><b>The Process</b></span><br />So, decision made, the first step was downloading <a href="http://www.microsoft.com/windows/windowsmedia/forpros/encoder/default.mspx">Windows Media Encoder 9</a>. After suffering through the hell that is the Windows Genuine Advantage Validation process, I finally downloaded and executed the file, installing it without problems.<br /><br /> <img src="http://www.digitalmediathoughts.com/images/WM_encoder.jpg" /> <br /><br />Next, I needed to configure Exact Audio Copy to use the WMA Lossless compression profile. If you’ve never used EAC before, I encourage you to try it. As I’ve said, it’s a very powerful CD ripping program that produces the best-sounding MP3s I’ve ever heard. I won’t go into configuring EAC for initial use here, though if enough people are interested, I suppose I could be convinced into writing a guide to that. :wink: Changing EAC from MP3 compression to WMA Lossless was as easy as going into the Compression Options dialog, selecting ‘Microsoft WMA9 Encoder’ as the Parameter passing scheme and ‘Lossless’ as the Bit rate.<br /><br /> <img src="http://www.digitalmediathoughts.com/images/EACWM9.jpg" /> <br /><br />It was so far, so good at this point, so I grabbed a CD off the ‘to be ripped’ pile and popped it in the drive. As usual, the CD contents were read and, after a brief consultation with freedb, displayed in the main EAC window. To my amusement, the CD I had grabbed happened to be almost exactly 60 minutes in length, which I took as a sign that my testing was going in the right direction. Crossing my fingers, I clicked the now-mislabeled ‘MP3’ button. The drive spun up and EAC began reading the first track. Thankfully, it was a short track and I didn’t have to wait long before EAC launched the WM9 Encoder in command line mode. I watched with a small sense of victory as the numbers went up to 100%, the box closed and EAC moved on to the next track. I sat back and relaxed while the entire CD was ripped to the hard drive.<br /><br /> <img src="http://www.digitalmediathoughts.com/images/WM9_encoding.jpg" /> <br /><br />A beep sounded when ripping was complete, and I went immediately into Windows Explorer to check out the results of my experiment. My initial reaction was shock at the massive file sizes, but I had somewhat expected that as I remembered FLAC files being quite large themselves. Then I realized that I needed comparison data, so I re-ripped the same CD into different folders as uncompressed .WAV and 256 kBit/s CBR MP3 files. My results were fascinating, to say the least.<br /><br /><PAGEBREAK><br /><b><span>The Results</span></b><br /> <img src="http://www.digitalmediathoughts.com/images/413MB.jpg" /> <br /><br />As I said, I was shocked, but not wholly surprised at the results of my tests. My experiment revealed that a 60-minute CD translated into 413MB worth of Lossless WMA files! To put that into perspective, the uncompressed .WAV files took up 614MB, whereas the MP3 files came out at 111MB. So the Lossless WMA files were almost four times the size of high-bitrate MP3s, and saved just a third of the file size over not compressing them at all. Hmm. Here’s that info put succinctly:<br /><br />File Type:......................Size:........Time to Rip:<br />Uncompressed .WAV......614MB.......~10 minutes<br />Lossless WMA................413MB.......11:34<br />256 kBit/s CBR MP3.......111MB.......12:34<br /><br />I now had definitive file size data to work with, and it was telling. What about the sound quality though? That’s what I’m really striving for here, otherwise why bother changing what I’m doing at all? So I closed up my office and cranked up the speakers to 11 and listened.<br /><br />I’ve stated that I’m not a true audiophile, but I do love music. I’ve always thought (and preached) that a high-bitrate MP3 was indistinguishable from a CD. Well, I’m here to say <i>I Was Wrong</i>. A 256 kBit/s MP3 is widely considered to be ‘CD-quality’, but I’ll be darned if I didn’t hear a difference between the two. The song I chose is one of my favorites, so I know it well, and it’s full of many rich independent sounds, so I figured it would be a good test. I skipped to a climactic point where I knew the band went all-out. I listened carefully to each type of file, back and forth. It was hard to tell, because I was listening to it much more intently and at a higher volume than I’d ever had before, but I can tell you for sure that at least one sound that I had always figured was noise (this band likes noise) was actually another instrument way in the background! To me this was definitive proof that an MP3, even a high-bitrate one, will throw away sound that might prevent you from properly feeling the full experience. And it proved that ripping CDs into a lossless format might be worthwhile after all.<br /><br />Once I had done that, I gave the format a more formal shakedown. It loaded into Nero just fine and it burned a copy with no problem. Theoretically, that burned CD should be a 100% exact duplicate of my original. That makes me happier than I expected, for some reason. Although there was no information on it one way or another, I was pretty certain my PMP wouldn’t play the lossless file, but I tried it anyway. Sure enough, I got a corrupted file message. Oh well.<br /><br /><b><span>What Does This Mean?</span></b><br />I now knew that WMA Lossless was clearly superior, sound-wise, to the MP3s I’d been ripping and listening to. I also knew that lossless files would take up almost four times the space of the MP3s I’d been listening to, and I knew that ripping a CD into WMA lossless was just as easy, and a tad faster, than what I was currently doing. Take into account that my current MP3 collection is 60GB, and my brain started spinning at the implications. I couldn’t just times 60GB by four and get the size of my collection as WMA lossless because I knew not all of those files were at 256 kBit/s quality. The more accurate estimate would come by multiplying the time factor. When I import my entire MP3 collection as a playlist, WinAmp reports I have just over 31 days of continuous music! Round down to 410MB per hour of music, and you get just over 305GB for 31 days of music.<br /><br />Now, I keep my music collection on my 500GB NAS, so 305GB is pretty pricey, but not unaffordable to me. It’s not just about the space though. 305GB represents a significant time investment on my part. In fact, at an 11:30 rip-time per 60 minutes of music, that’s over 142 hours (almost 6 days) of straight ripping with no breaks. Heh. Needless to say it would take me considerably longer than that to do it.<br /><br />In addition, there’s the fact that my new hypothetical lossless collection would not be portable. I would have to convert every file I want to carry on my PMP. Talk about a drag. I’m somewhat used to that though; I already have to convert any videos I want to watch on it.<br /><br /><b><span>My Conclusions</span></b><br />So I’ve decided to do it. I’ll take the plunge. From this day forward, I hereby pledge to rip every new CD to WMA Lossless. I do pledge to slowly, one by one, convert all of my current MP3 collection to WMA Lossless by re-ripping every CD I own, starting with the ones we listen to most often. I’ve decided I’ll store my massive new lossless collection on my NAS and store the albums I want to be portable locally on my main machine’s hard drive. I will keep my current MP3s of those albums as I re-rip them, and convert any new ones I want. I used <a href="http://www.dbpoweramp.com/dmc.htm">Illustrate’s dBpowerAMP Music Converter</a>, with its <a href="http://www.dbpoweramp.com/codec-central-wma.htm">WMA codec plugin</a>, to convert my lossless WMAs to MP3s in no time with a simple right-click. For $14USD, its value to me is considerable.<br /><br />It’s a huge task and a huge commitment, but I think I’m up to it. I only wish I’d done this years ago. Get back to me in 2010 and I’ll let you know how it’s progressing. ;)<br /><br /><i>Damion Chaplin is a graphic artist and digital media connoisseur who’s secret identity is a mild-mannered internet manager at a hardware store in San Francisco.</i>

[Jason Dunn]

A very interesting article! I've been thinking about going lossless myself, but your article convinced me to do a comparison test between what I normally rip in (256 kbps MP3) and WMA Lossless. I used a Jars of Clay song, World's Apart, because I know it so well. I listened on my Logitech 5.1 speakers, and I honestly couldn't tell the difference. The beginning of the song has wind chimes, which are definitely in the high-end range of what usually suffers from compression artifacts. I couldn't tell the difference. I also listened for subtle things, such as the sound of the guitar player moving his fingers on the fretboard as he changed chords, and both files had the sound, and neither one sounded better than the other.

One song does not a conclusion make - I'll repeat these tests again with a different song - but for me at least, 256 kbps seems to be enough.

[Damion Chaplin]

...and neither one sounded better than the other.

Well, frankly, I probably wouldn't be able to tell the difference on 99% of the music I own. I just happened to luck out in the song I chose I guess. Now it's more a matter of principle to me, not to mention I'm just anal and need my entire collection to match. Now I really do have exact audio copies.

And in case anyone's interested, the song I used to test with was The Mars Volta's Drunkship of Lanterns, the phrase beginning at 4:20 on the counter. 8)

[Janak Parekh]

Yup, interesting article. Nice observations, Damion. :)

Interesting footnote: Apple is "ahead" of the curve here, as Apple Lossless is supported on iPods, but there is (obviously) a serious hit to battery life as far less music can be cached. For what I use my music collection (on-the-go and casual office listening), 160kbps is enough; I can notice the artifacts, but the sound system is generally not good enough to bother at a higher bitrate, and I don't want to carry a large external disk with me.

To someone like me, the more important value proposition would be format independence; I could reencode a lossless track into whichever format I wanted for whichever player I want to carry, thereby avoiding WMA or AAC lockin. Still, I don't have the energy to do this at this point. ;)

--janak

[Janak Parekh]

...and neither one sounded better than the other.
Well, frankly, I probably wouldn't be able to tell the difference on 99% of the music I own. I just happened to luck out in the song I chose I guess.
It should depend on the music. I do have the first Jars of Clay album, and I haven't listened to it recently, but if I remember correctly there are only 2-3 active instruments at any given time in most of the songs. Something like classical music, where there could be an entire orchestra playing -- and more high-frequency instruments, like strings, cymbals, etc -- should in theory yield more noticeable differences.

--janak

[Jason Dunn]

To someone like me, the more important value proposition would be format independence; I could reencode a lossless track into whichever format I wanted for whichever player I want to carry, thereby avoiding WMA or AAC lockin.

Yes, same here, that's something that I'd value as well. Though 256 kbps CBR MP3s are compatible with EVERYTHING AFAIK, so I don't feel very locked in. ;-)

[Doug Johnson]

I'd be very curious to hear your take on WMA VBR. It is a much better sounding format than MP3. Expecially in its latest incarnations.

I did my own experiment a while ago (about 18 months I'd guess?) where I ripped several different genres of music into WMA, MP3, and WAV, at several bitrates, using both CBR and VBR. I invited a few friends over and played samples of each file on a high-end audio system.

The result? The MP3, even at high bitrates, always stood out compared to the others, as being harsher, tinnier, and rough on the high end. And there were also descriptions of it not sounding as big or spacious (less out-of-phase material?) Everyone could hear it, even though all but one had untrained ears. The lower bitrate (&lt; about 128kbps) WMAs could be distinguished from the WAV -- described as "missing something" or "dull," but higher bitrates could not be by most of the listeners.

MP3 is a much older format that has not evolved over time, so its technology is quite dated now (think "late 1980s/early 1990s"), whereas WMA is newer and has been kept relatively up-to-date. Another thing I have noticed is that WMA encodes a lot faster than MP3 (at least the LAME encoder, which is what I have used.) The downside of WMA is compatibiliy... it hasn't been supported by all audio devices, such as my car stereo, which only supports MP3. And support in other OSs is limited. But in terms of actual sound quality, WMA appears to be much better than MP3. Something to consider when ripping your CDs.

[Janak Parekh]

The MP3, even at high bitrates, always stood out compared to the others, as being harsher, tinnier, and rough on the high end.
What was the highest bitrate you tested? From what I've heard, MP3 is near-indistinguishable at 256kbps and better, and that WMA, Ogg, AAC are all particularly optimized for lower bitrates (not that they would do worse at 256kbps, but not noticeably better).

I'd agree that there's certainly a difference at 192kbps or lower.

--janak

[BugDude10]

Damion, outstanding and informative article! Thank you!

I've been perfectly pleased with 160kbps WMA files on my portables (never used for serious listening -- that would be original CDs on the home stereo system), but I've wondered about whether lossless formats would be worthwhile. I don't have the drive space for my entire collection at 160kbps (over 50GB), so WMA lossless wouldn't work; maybe with a bigger drive and a quick &amp; dirty way to convert them into something that my portables can handle. (My portables are all iRivers, except for my HP Pocket PC and my PSP, and iRiver has always been good at multi-format support... perhaps lossless is in the future somewhere...)

I'm probably not gonna jump into a lossless re-ripping festival any time soon, though -- I just got my collection re-ripped at 160kbps a few months ago. Not a project I'd look forward to again!

Once again, thanks for the information!

[Damion Chaplin]

...The downside of WMA is compatibiliy... it hasn't been supported by all audio devices, such as my car stereo, which only supports MP3. And support in other OSs is limited.

This pretty much answers your question I think. Originally, I went with MP3 because it was pretty much the only thing out there. WMA was a brand new format and the only thing that read them were my iPaq. So all CDs were encoded in MP3 format. I went for CBR over VBR for pretty much the same reason: My MP3 player at the time only supported CBR. So CBR MP3s were by far the most cross-compatible format there was - and still is I think.

That being said, during my testing I did rip into 256 CBR WMA and the file results were almost identical - 112MB. That's actually surprising because I remember 128 kBit/s WMAs being about 2/3 the size of their counterpart MP3 files, but I didn't stop to ponder it. I didn't concentrate on the sound either as I wasn't doing that particular comparison at the time, but they pretty much sounded the same to me.

VBR files will, in general, be both smaller and sound better than CBR files. That's more or less a given. As I said though, I've always wanted to guarantee compatibility, and that means 256 kBit/s MP3s. At least for my local, portable collection. :wink:

[Jason Dunn]

That being said, during my testing I did rip into 256 CBR WMA and the file results were almost identical - 112MB. That's actually surprising because I remember 128 kBit/s WMAs being about 2/3 the size of their counterpart MP3 files, but I didn't stop to ponder it.

No no no - but a common misconception. :wink: I've written an article about this very thing, to be posted soon (though not here). A 256 kbps WMA or MP3 is going to be the same size (give or take a few KB) because the data in the file is just that: 256 kb of data per second of audio. The difference is that WMA *sounds* better at a lower bitrate.

[Doug Johnson]

What was the highest bitrate you tested?
320kbps. It was definitely better, but still wasn't quite transparent.

[Doug Johnson]

Originally, I went with MP3 because it was pretty much the only thing out there. WMA was a brand new format and the only thing that read them were my iPaq. So all CDs were encoded in MP3 format...
I was curious if you compared WMA VBR vs WMA Lossless/WAV in your testing, from a sound quality standpoint. Curious if you could hear a difference.

[Damion Chaplin]

The difference is that WMA *sounds* better at a lower bitrate.

Wait, so I'm not remembering it correctly? The WMAs I was using in days of yore weren't actually smaller than the same bitrate MP3s? I sure remember them being so, but you know memory is. :roll:

I was curious if you compared WMA VBR vs WMA Lossless/WAV in your testing, from a sound quality standpoint. Curious if you could hear a difference.

No, I didn't perform that test. Once I had confirmed that any compressed format (at a high bitrate) does throw away actual sound, that was more or less the end of my love for any said compressed format. It's a good point though and I'm curious myself. I'll check it out and let you know.

[mcsouth]

I've done my own experimenting over the last few years, with the result that I am ripping CD's to WMA Lossless on my PC, but I maintain a library of lower bitrate 'duplicates' of some music (not all by far) on another drive - mainly so that WMP doesn't try to catalog it, showing duplicates in my Library. So far, I have about 100 hours of lossless music, taking up about 35GB of disc space, so obviously, my collection is not as extensive as some.

A lot of my low bitrate music is 64 to 128 kbps WMA's, which were definitely better sounding than MP3's at the same bitrate. I didn't bother encoding at a higher rate, mainly because of capacity restrictions on my PDA and MP3 player (512MB in PDA, 256MB player), and the fact that cheap earbuds weren't going to do their part anyways with higher quality material.

One thing I have been doing more frequently now is building playlists of music; I will periodically move the files off my MP3 player to this other drive, and then have WMP sync a new playlist of content that I don't have low bitrate versions of - sync process is set to reencode at 64 kbps currently. It takes a while, but I just get it started, and then wander away to do other things. (Thanks to Jason for pointing out this WMP10 feature to me some time ago!)

[Felix Torres]

I was curious if you compared WMA VBR vs WMA Lossless/WAV in your testing, from a sound quality standpoint. Curious if you could hear a difference.

No, I didn't perform that test. Once I had confirmed that any compressed format (at a high bitrate) does throw away actual sound, that was more or less the end of my love for any said compressed format. It's a good point though and I'm curious myself. I'll check it out and let you know.

Please do. Would be an interesting data point.

A friend of mine who is very much into classical music, among other things musical, found high bit-rate wma vbrs to be *truly* equivalent in sound to lossless and (as you pointed out) smaller than the equivalent cbr's. By a lot. So a 320kbps wma vbr may be smaller than even a 192 cbr, but sound as good as lossless.

I need to run this same kind of testing (I've only done a few cds worth so far, so I can't fully confirm this but so far it seems to be true) but my main focus at this point is XBOX360 compatibility; I want to see which is the best sounding format I can play off a USB mass-storage device (external hard drive, actually) via 360. (I'm thinking of getting my mother an XBOX360 for XMAS. :twisted: )

As is, the 360 plays lossless but...not...quite...right...
Need to see if its because I was feeding it off a flash card (worked fine for compressed files), because I fed it a bad file (it played fine on a PC, though) or if the 360 Media player doesn't really like lossless (possible...maybe even probable. But then, the Media Player is being update reall.soon.now.)

[klinux]

A friend of mine..

Audio perception is such a qualitative and personal preference that such statement is analogous to a friend of mine likes blue so you may like it too.

So a 320kbps wma vbr may be smaller than even a 192 cbr

No it won't; Jason just discussed this. A 320kbps WMA VBR files contains 320 kilobits of data per second and should be 66% bigger (plus or minus a few %) than a 192 kbps WMA/MP3/AAC/whatever CBR because it literally has 66% more 0's and 1's!

My personal strategy is similiar to that of mcsouth. I rip to a lossless - any will do since in theory the various lossless formats are same but differs in size, which is cheap, and CPU usage, which is also cheap. In my case, I use Apple lossless. For portable or cross platform (ditto Doug Johnson) listening, I mix lossless or lossy (mostly MP3) depends what I have available but I agree completely with mcsouth that headphones and less than ideal listening environments make the quality issue moot for all but the most discerning listener.

My music preference is classical and vocal. I listen on my Magnepan speakers when possible.

[Felix Torres]

So a 320kbps wma vbr may be smaller than even a 192 cbr

No it won't; Jason just discussed this. A 320kbps WMA VBR files contains 320 kilobits of data per second and should be 66% bigger (plus or minus a few %) than a 192 kbps WMA/MP3/AAC/whatever CBR because it literally has 66% more 0's and 1's!


Uh, sorry to burst your gotcha moment but that's only true for cbr files, not vbr files.

Sorry if my use of acronyms is confusing you

CBR = constant bit-rate and yes, file size is fixed (more of less) by the bit rate.

VBR = Variable Bit rate, which means exactly that; the bit rate *varies* from instant to instant and the reported bit-rate is merely an estimate of the usage space. Depending on the type of music the compression effect could be (and often is) much higher than for a nominally equivalent CBR file.

As for my quoting classical music perception; well, classical is the most demanding music out there for this kind of exercise because of its extreme dynamic range requirements. The effect Mr Chaplin noted with "new" instruments appearing out of the "noise" under lossless is more likely to appear on a heavily orchestrated classical overture than it would be in a banjo solo or rap piece.

Sorry if I confused anybody.

[Damion Chaplin]

No it won't; Jason just discussed this. A 320kbps WMA VBR files contains 320 kilobits of data per second and should be 66% bigger (plus or minus a few %) than a 192 kbps WMA/MP3/AAC/whatever CBR because it literally has 66% more 0's and 1's!

Actually, I don't believe that's true. My understanding is that a 320 kBit/s VBR file is a variable bit rate up to a maximum of 320 kBit/s. During quiet periods, or periods of silence, the bitrate in a VBR file drops to next to nothing. Only during the most cacophanous parts of the song will the bit rate actually reach 320 kBit/s, whereas a CBR file keeps the bit rate at a constant 320 kBit/s, regardless of what noise data is actually playing. That's why a VBR file of the same bit rate will be considerably smaller than its CBR counterpart.

I could be totally wrong though. :)

[Damion Chaplin]

So, I got all my music in Apple Lossless format for my PC and iPod, which I ripped from my CD collection. Any way to batch process them to 128kbp/s for my Microsoft Smartphone?

dBpowerAMP (mentioned at the end of the article) will convert Apple Lossless to MP3 with the Apple AAC plugin. I know it's capable of scripting, so you could probably batch files that way. I usually just go to the folder where the songs are, select all, right-click, convert.

[Janak Parekh]

Uh, sorry to burst your gotcha moment but that's only true for cbr files, not vbr files.
Not really, from my experience. VBR files may sometimes be a tad smaller, but it's on the order of the same size as the corresponding CBR bitrate -- the point is more to allow higher than the CBR rate during complex parts, and to make up the extra space overhead by using lower than the CBR rate during simpler parts.

It does get more complex, as there are multiple VBR algorithms, but that's generally the case, and my anecdotal data back when I experimented with this extensively suggests that VBR files can be both smaller and larger than the corresponding CBR files. The space savings can be accomplished because, for example, a 192kbps VBR may sound at about the same quality as a 256kbps CBR.

As for my quoting classical music perception; well, classical is the most demanding music out there for this kind of exercise because of its extreme dynamic range requirements. The effect Mr Chaplin noted with "new" instruments appearing out of the "noise" under lossless is more likely to appear on a heavily orchestrated classical overture than it would be in a banjo solo or rap piece.
Agreed, as I mentioned earlier. ;)

--janak

[Janak Parekh]

My understanding is that a 320 kBit/s VBR file is a variable bit rate up to a maximum of 320 kBit/s.
I'm not 100% sure about WMA as I ditched it years ago, but I'm certain that this is not the case with MP3. I used to rip extensively to MP3 VBR, and one of the great features of the old version of Winamp is that during playback, it would show you the "current" bitrate in real-time. It would definitely go both below and above the targeted average bitrate.

Now, to complicate things, 320kbps may be a special case. Many compressed audio formats have a maximum bitrate; if a format's maximum bitrate is 320kbps, then VBR would indeed act as a "lossy" compressor on top of the codec itself. But that's not the original point of VBR, just an interesting side-effect.

--janak

[Jason Dunn]

It does get more complex, as there are multiple VBR algorithms, but that's generally the case, and my anecdotal data back when I experimented with this extensively suggests that VBR files can be both smaller and larger than the corresponding CBR files. The space savings can be accomplished because, for example, a 192kbps VBR may sound at about the same quality as a 256kbps CBR.

Well, it really depends...there are a lot of variables here. If you compare a VBR file capped at 256 kbps to a CBR 256 kbps, the VBR will be smaller. If that same VBR is capped at 320 kbps, it will likely be bigger, but still smaller than the same MP3 at 320 kbps CBR.

So delightfully confusing! :lol:

[Janak Parekh]

If you compare a VBR file capped at 256 kbps to a CBR 256 kbps, the VBR will be smaller.
So, I haven't ripped with WMP in ages, and I notice you use the word capped. If that's indeed how WMA VBR works, it's different than many MP3 VBR encoders which don't cap, but rather average at that point.

--janak

[Jason Dunn]

If that's indeed how WMA VBR works, it's different than many MP3 VBR encoders which don't cap, but rather average at that point.

Yeah, I think this has changed over the years (to further complicate things) - I know for sure when I did my original VBR rips years ago in MP3 you could specify a ceiling. Now with LAME you specify a general target point, but it can go above it.

So, basically, when you're talking about VBR, there are no hard and fast rules. ;-)

[mcsouth]

Just out of curiousity, does anyone here have a feel for how many current MP3 players can actually handle VBR files? If I recall when I reviewed this issue a year or two back, that was one of the issues - not all players could handle VBR files, or handle them properly. That was one of my motivations to go to lossless - high quality on the desktop, but I had the option to downsample to a format that a particular player would handle.

Depending on the vendor's site, or even manufacturer's packaging, there often isn't any specific mention of VBR v. CBR file support.

[Jason Dunn]

Just out of curiousity, does anyone here have a feel for how many current MP3 players can actually handle VBR files?

A couple of years ago, it was hit and miss. Nowadays, I think most do, but because of what I went through a couple of years ago with VBR, I only rip CBR now. The potential hassle just isn't worth it.

[klinux]

Uh, sorry to burst your gotcha moment but that's only true for cbr files, not vbr files.

Sorry if my use of acronyms is confusing you.

No, you did not confuse me at all. I know what VBR and CBR means. The fact that I explicity mentioned "plus or minus a few %" should be a dead give away. In addition, implicitly this is also saying that some encoders use aim for a ceiling, some aim of minimum, and some average. This should have been explicitly stated.

Nevertheless, I still stand absolutely by what I said. Here is a real world challenge: in your own CD collection, you will not find one single album that will give you a smaller file size encoded in 320kbps VBR in whatever codec than using 192kbps CBR using that same codec .

During quiet periods, or periods of silence, the bitrate in a VBR file drops to next to nothing.

Quiet periods does not equal to silence! I do not think this is true and would like any one that can substantiate that statement.

Lastly, any classical music listener, like myself, can tell you duh, we obviously know that our music is the most demanding than heavily compressed synthesized pop. But the music codec that is right for us - lossless - may not be right for you, your music, your listening environment, your equipment, and so on. Decide for oneself based on one's individual conditions.

[sapibobo]

Maybe it is slightly out of topic. But i tried EAC yesterday but ends up in large .wav files only.

I set up compression option to WM9 encoder but nothing happened. EAC just create those large .wav files.

Any guide? I red the developer web site but i get confused with the settings, especially which .dll file goes which folders etc...

Anyone kind enough to help me with step by step guide? Or any links or resource that can clearly explain what to do?

TIA

[Damion Chaplin]

Anyone kind enough to help me with step by step guide? Or any links or resource that can clearly explain what to do?

Hey sapibobo and welcome.

I think we'll be posting an article on that very thing in the pretty near future. Stay tuned! :)

[Lee Yuan Sheng]

I listen to classical in 192kbps VBR. :P Does that make me an idiot or a heretic? :P

[firedog]

Hi Guys-

I have extensive experience with lossless. I've used monkeys audio and FLAC extensively, as well as WMA. In fact, I've encoded my entire music collection to FLAC (about 300 CDs) and have a quiet computer just for audio hooked up to my stereo. I barely use my actual CD's anymore, as I have the entire collection (including features such as playlists) available at the click of a mouse. The actual Cds are now my "backup".

There is no question that lossless formats sound better. It's just that you need a good stereo and good ears to tell the difference.

If you're listening on a portable device, or a compact stereo I doubt you will hear the difference at least at first. You need at least a good consumer stereo (ballpark guess: $500 minimum, probably more lke $750 or $1000) with good speakers (or an audiophile rig) to clearly hear the difference.

In addition, if you're used to listening on lesser equipment, it takes time for your ears (actually your brain) to adjust when you switch to better equipment. You won't notice a difference at first, but after a few days of listening to better stuff, if you go back and listen to lesser equipment you will wonder how you ever thought it sounded good. This is a known phenom in the audio world.

I have a true audiophile rig and the difference between compressed and lossless formats is instantly noticeable and very significant. It is most noticeable in the very high end (e.g. cymbals) and deep bass. They simply sound crisper (cymbals) and more solid (bass). But the overall sound is also much improved: more "space" between instruments; better stereo image; more detail in the music. You do actually hear things you don't hear in the lossy formats

The lossless formats are very good. I think they sound exactly like the original CD. They each have slight differences. I think Momeys Audio is the best at high compression (you can compress to about 45-50% of the original size of the wav file); but overall I like FLAC better at somewhat less compression. Presently my entire music collection is compressed into FLAC. I think it is about 40% smaller than it would be in wav. It takes up about 130GB. FLAC is also more widely supported than Monkeys, and I think over time it will gain support.

You can download free codecs for all these formats and use Winamp to or some other players to play them. Several ripper programs such as EAC or various Lame Front end programs will accept the codecs and rip to the formats. A simple Google search and you can find whatever you need. I highly reccomend using them if you have disk space.

Anyway, that's my two cents.

...I'm now editing this post and adding an additional paragraphs, since I think the post came off harsher than I meant it. I wasn't trying to say there is anything wrong with lossy formats. I also have an mp3 player and use mp3's, as they make sense when your storage space is limited. They also sound good when played back through ear pods or moderately priced equipment.

I also wasn't trying to imply that there is something wrong with a reasonably priced consumer sound system. My point was simply that mp3 isn't really "CD Quality". Even at 320 kbs, there is a noticeable difference in quality, if you know what to listen for and/or have equipement that can playback the difference. Moderately priced equipment can't really do it. It's not intended to.

That's why people like me spend lots of money for expensive equipment that brings out the true sounds of the recordings, even though the sound improvement per dollar ratio decreases as more money is spent.

I do, by the way, think that many of you could hear the difference in the lossy and lossless formats on your own equipment with a bit of practice. It just isn't obvious when you first try it. The more you listen, the more the differences start to be noticed.


Danny

[Jason Dunn]

Thanks for taking the time to share your thoughts Firedog (Danny)! It's always interesting to hear another opinion. I don't own any audiophile-class speakers, all consumer-grade stuff, so I guess I'm safe with my compressed audio. ;-)