Slashdot Mirror
Ogg Vorbis RC3 Released
xercist writes: "Let's start 2002 off with some good news! The long awaited RC3 release of the Ogg project's Vorbis codec is now out. Major changes include much improvement in the quality to bitrate ratio, ability to specify a hard bitrate min/max to the encoder (good for streaming), and an entirely new bitrate management engine which can emulate CBR, do constrained bitrates, and will accept quality settings via the -q flag from 0 through 10 in .00000001 increments (currently only tuned for 44.1 KHz modes). Vorbis has kicked MP3's, WMA's, and Real's asses for a long time now, hopefully this release will change the minds of anyone yet undecided. Download RC3, then show your appreciation for all their hard work and dedication by making a donation to support the project."
If you want to take the listening test yourself, read the instructions and jump in. For now, there's also a page of interim results, but to quote ff123, "Major conclusion: I need more listeners!"
Monty
The best tests we have at the moment were conducted by
ff123 at 128kpbs. There have been two so far (the second is technically still underway, although it's now based on outdated encoders, so I imagine a third will start fairly soon). The
first listening test compared RC2 Ogg Vorbis, LAME MP3, Xing MP3, Liquifier AAC, MPC, and WMA8. The formal analysis showed that, on the file compared, the encoders could be divided with 95% confidence into three groups (from best to worst):
- MPC and AAC
- WMA8 and LAME and OGG
- XING
.The second test used a CVS version of OGG from about a month and a half ago. This time there are three test samples which participants can choose to evaluate. While technically still underway,
the interim results can be found here. Of the three test samples, the first can't discriminate between the encoders, the second looks like it will but needs more listeners (and the results so far look interesting), and the third discriminates well, to the extent that it shows that Xing and WMA8 are statistically much worse on that clip than all the others.
Now all we need is a third test with the latest updates of all the encoders - since we now have a new stable version both of Ogg Vorbis (RC3) and LAME (3.91).
-- Help Digitise the Public Domain at DP.
Two thoughts:
1. NEVER EVER CONVERT BETWEEN LOSSY FORMATS, it will add unnesessary artifacts and ruin the audio quality.
2. I wasn't aware that the thought police would be any more able to charge money for posession of MP3's than Ogg's .
RC3 is much better quality all round than RC2 (and RC4 will be better quality than RC3 particularly for low bitrates). Particularly, RC2 had some problems with high frequencies that have now been fixed.
So reencode - and use '-q' switches this time instead of '-b' switches...
-- Help Digitise the Public Domain at DP.
There are plans afoot.
The Ogg Vorbis *decoder* has been stable since RC1, and will be able to play any Vorbis stream produced by RC2, RC3, 1.0, or whatever. There are slight problems in that the reference decoder is floating point, which doesn't fit well with the ARM chips a lot of hardware players use, but that'll be sorted eventually.
-- Help Digitise the Public Domain at DP.
Monty
Rio doesn't have a general purpose chip. They have hardware for mp3 decoding. There is no hope for Ogg in most Rio devices. Sorry.
The problem is not the AVI or Ogg; it's that the vast majority of AVI *players* cannot handle a VBR codec. These players ignore all the sync timestamping, assuming the audio is coming in CBR.
Have a look at ww.hydrogenaudio.org for discussion of players that work properly.
Monty
It's not 128 Kbps MP3, it's 256 Kbps MP3. I can consistently tell 128 Kbps MP3 from the original rip, even on cheap $15 multimedia speakers (although I have to hold them right up to my ears). And I'm no audiophile.
Go to http://www.r3mix.net/ and click on the "Quality" link for some links to the MP3 tests.
For chatting with developers real-time (but no guarantee when we'll be there), catch us on #vorbis at irc.openprojects.net.
Monty
With new computer hardware coming my way I've decided it's time to abandon the mp3 format in favor of Ogg. And thnkfully I've already got the tools on hand to make it easier.
http://admin.dbpoweramp.com/
db Power AMP is a handy tool I found in my quest for auto-mated audio conversion. I had a bunch of ASF files (no video) that I wanted to convert to Mp3. And I sure as hell didn't want to do it by hand, extracting a wav then reencoding it. So I went hnting and found that program. It doesn't do directories, but it DOES do playlists.
All I gotta do is wait for an Ogg made for db Power AMP and also wait for my new motherboard to arrive.
I'll be happily moving to a Tyan Thunder K7, with fond memories of my Abit KT7-RAID. But my old board will stick around long enough to play work horse.
I'll make a playlist in winamp with all 24 and a half gigabytes of mp3's, and let the box sit quietly while db Power AMP converts from mp3 to Ogg.
Oh! It does any and all available formats by the way, well practically all of them. you'll need specific codecs for conversion. They don't have one for RC3 yet. But I'm sure it's on the way.
Computational Madness in a round package.
I'd write that in normal conditions, but I believe you'll use that to convert MP3 to Ogg.
You shouldn't do that, because you'll get artifacts from MP3, artifacts from Ogg and extra artifacts added and the resulting file will sound crappy.
So unless you give me another reason to write such an app, I won't.
Because vorbis.com is becoming slow, I have decided to post mirrors:
win32 binaries: vorbis-tools-1.0rc3-win32.zip
i386 RPM libao: libao-0.8.2-1.i386.rpm
i386 RPM libogg: libogg-1.0rc3-1.i386.rpm
i386 RPM libvorbis: libvorbis-1.0rc3-1.i386.rpm
i386 RPM vorbis-tools: vorbis-tools-1.0rc3-1.i386.rpm
To encode files, you need all the above RPMs.
There's little question that Vorbis is impressive. The question is, what is its competition? MP3 (created using LAME) is currently the most popular digital audio compression algorithm, but anyone will tell you Vorbis rocks its world. That can't be it, then... is RealAudio/WMA the true competition? How about Quicktime? Perhaps Vorbis is playing to different audience than the "big boys," mainly for the home enthusiast? Vorbis is not quite ready for streaming (e.g., not yet perfectly tuned for 22.1kHz like for 44.1kHz, not very low bitrates, etc.), so until then it seems Real will lead the pack in that arena.
When, however, Vorbis gets these features, I feel it will even be able to replace Real and WMA.
About once every... oh, 10 minutes... someone asks for a tool to convert MP3 to Ogg.
Do NOT convert MP3 to Ogg! Converting (transcoding) between lossy codecs only makes the quality horrible -- the artifacts interact in unpredictable ways. It's like faxing a photocopy of a fax.
Rip your CDs with Exact Audio Copy (win32) or cdparanoia (Linux, et al.). Encode them with oggenc (or LAME if you need MP3 for portable devices). Share them with your friends.
As for ABX: Oops, you're right. The results ff123 asks for are not ABX, they're the traditional 1-5 scale that MPEG has always used. ff123 *does* suggest using ABX to certify the results, but that's not the same thing, and you're right to point that out.
Last, parts of the tests are automated, parts aren't; if you go the ABX route, there are automated testing packages to use (linked from ff123's page). I've not added my results to this test only because it's a little too easy for me to cheat. So, I didn't go through the test process myself, I've only been watching the results.
Monty
This is not a slight problem, this is the problem for ogg on hardware players. You will never see ogg on anything other than your pc until this gets done.
While I can't vouch for the iPod or iTunes (sadly), I can make a comment about QuickTime: one group of developers has finally written an Ogg QT plugin that allows you to use Vorbis audio in QuickTime just like everything else. That's definitely a step in the right direction. It should allow you to play Oggs from within QT. If iTunes can play any QT-supported audio codec (and quite frankly I simply cannot remember whether or not it relies on QuickTime; sorry), then you now have the ability to play Oggs from within iTunes if you wish.
I know everyone will get into a discussion about music quality... so here's another question.
.. they were about the same price as my old but trustworthy Sennheiser HD330s.
We all know (I hope) that what you hear is also limited by your listening equipment.
I recently bought a pair of Sony MDR-V500 headphones
I was dissapointed when I actually had them side by side; the Sony headphones are basically, well, crap. Any listener could distinguish that they are severely lacking in several areas. The sennheisers sound oh so much better.. and that's on a computer, through a cheap desktop speaker headphone jack, listening to 160Kbps mp3.
So what's the point of arguing over compression formats, or whether something is *really* CD quality, or studio quality, when your equipment can't even come close to reproducing it?
Oh.. to the unitiated.. I highly recommend a good pair of $100 headphones (Sennheiser or Grado, and yes, that means towards the lower end of their product lineups..don't let that discourage you. A low-end Grado or Sennheiser sounds fantastic compared to anything else you'll find in the store.
And those $100 headphones will sound better than a $2000 stereo, anyday.
So what do you guys/gals use?
Someone contacted iRiver (the Korean OEM who *actually* makes the Rio Volt) and they said it was possible. Supposedly they're waiting for the Magic Version 1.0 before implementing it.
"A few weeks later he upgraded his operating system, and WMA's Rights Management kicked in and told him he couldn't play any of those files anymore. "
That would explain why I turn of Digital Rights Management in Windows Media Player.
I believe Microsoft also now provides a backup system for the DRM stuff. Haven't tried that, again because none of my music uses DRM.
Two things.
First - you say it's a drawback that Ogg at 117kpbs sounds better than MP3 at 128kpbs? I don't understand your point.
Second - LAME's legal status is unclear. The MP3 patents are not (only) in psychoacoustics, they affect the basic MP3 file format, together with natural (obvious) optimizations you would use in creating an MP3. LAME *does* infringe these patents.
-- Help Digitise the Public Domain at DP.
Yes - if you want to support Ogg Vorbis then *don't transcode* - Re-rip to Ogg. Otherwise all the file sharing networks will be flooded with poor quality Oggs, and we'll be no better off than we were with MP3s.
-- Help Digitise the Public Domain at DP.
One thing to remember -- vendors of embedded hardware doing audio recording and playback, commercial software with need for an audio format (ie. games w/ theme music) and the like need a good audio codec they can use without dealing with licenses or patents.
MP3 isn't this.
Vorbis is.
Harmful or not, it bugs me the way they use the term "Release Candidate". I don't know about the rest of you, but on every project I've ever worked on, once you start calling things Release Candidates, you stop adding features and just fix bugs until you get the actual release version nailed. Generally speaking, your first RC should be something you think has a shot at becoming the actual release, otherwise why are you calling it a candidate? This is obviously not the case with Vorbis, as they keep adding new stuff with every candidate.
Now, having said that, there's two things I'll point out. First, I can't find anywhere where they actually spell out "Release Candidate" so maybe RC actually stands for something else. Second, this post (mine, not the one I'm responding to) boils down pointless nitpicking over semantics. They can call it whatever makes 'em happy. But following the same conventions as the rest of the world makes it easier to figure out what they really mean. In the meantime, I will just mentally replace "RC" with "Beta" when reading about Ogg Vorbis.
Perhaps another way to get people to start using Ogg Vorbis would be a better way to rip CDs into .ogg files. I've helped a few friends who aren't as into computers get Exact Audio Copy and LAME working together, and its not exactly the easiest process - people don't want to have to configure things for 10 minutes before they can start using the program, not to mention having to search for precompiled LAME binaries and then even after it's set up, having to tie up most of their computer's resources for the next 10 minutes while they rip/encode the cd. If Ogg had a very fast and easy all-in-one way for novices to do this (perhaps along the lines of iTunes, from what I've been hearing) more mainstream computer users might start encoding files that way.
The first ever Ultimate Frisbee video game: here (now
OK, I have to ask... why do people feel the overwhelming need to pontificate/ask profoud questions when they haven't even read the manpage? I'll summarize Ogg's VBR support. You'd have learned this from the FAQ, the READMEs, the manpage a trivial search of the mailing lists, or any of the previous Slashdot stories:
Ogg is natively VBR. It always has been. It's VBR is much better than LAME's because the format *itself* is natively VBR, not supporting it as an extra-spec hack that someone saw fit to kludge in later. Ogg's VBR output is and will likely always be higher quality than its bitrate managed (ie, ABR/BBR/CBR) modes. Don't use -b, -M, -m unless you actually have a *reason* to (eg, streaming). -q will always produce better results for the same output size.
[for the record, the following bits don't apply to this gentle poster, but to other comments]
Also to those below who are complaining, "wah, I reencoded my mp3s to ogg and they got bigger and sound worse," well, think for a second about what you've done. You've taken a lossy format, full of artifacts, and full of characteristics/artifacts specific to mp3 encoding. You're then encoding them in *another* lossy format, with it's own characteristics and saying 'do a good job'. Ogg is going to waste bits trying to reproduce mp3 artifacts perfectly. And because both formats are lossy (even if Ogg is very good), you still lose a bit in the process, a bit like transferring a cassette tape to 1/4" reel-to-reel. The reel to reel is pretty sweet.... but it's still a generational loss.
It seems exceptionally important to nip a few myths here. Most of you will laugh, but there are folks out there who still take a few of these as gospel, because sombody on some website four years ago swore up and down it was true:
- Decoding your mp3 to WAV and burning a CD does *not* improve or recover the lost sound quality. Once it was in mp3, those bits are gone forever. Similarly, converting from mp3 to ogg can *only* make it worse. It will not magically restore anything lost in the sound.
- bitrate is a measure of *size*, not quality. '128kbps' means absolutely nothing about file quality, just how big the file is. If you're rencoding mp3 into ogg (like a large number of folks here are...), of *course* making 256kbps oggs from 128kbps mp3s is going to result in bigger files! The encoder is doing exactly what you told it to.
- "VBR sucks. It saves space, but it's low quality and it messes up players." No, Xing's VBR mode sucks, and since they were the first mp3 encoder to hack this little travesty into a format that can't really support it, breaking most existing players at the time, people only remember Xing. Also add to this that Xing is consistently rated as the lowest quality of all commercial mp3 encoders, people who stopped learning in 1998 remember VBR as being a bad thing.
MontyIn Ogg, VBR is not a hack, it's native. We've been designing it that way for eight years. *VBR modes always sound better. Use them.*