Ogg The Conqueror? RC2 Is Out 329
jonathan_ingram writes: "There has been a lot of discussion recently in Slashdot about sound compression formats. Much has been focused on
Ogg Vorbis, but the most recent version available has been a beta released in Feburary.
Today, RC2 of Vorbis
has been released. The most important of the
many changes
is
channel coupling,
which means that Vorbis can now encode bitsteams at a much lower
bitrate than before.
Try it out today!"
Re:Dolby Digital (Score:1, Informative)
Re:WMP OGG Codec (Score:1, Informative)
Ogg V in the WSJ - (Score:2, Informative)
Don't know if it's online, too, but there's a nice piece of page 1, Section B on Ogg V. and C. Montgomerey.
I've got a mirror: (Score:1, Informative)
Re:Ogg Vorbis Quality (Score:3, Informative)
FYI, I have been using this for my MP3 jukebox:
webplay.sourceforge.net
I looked at a couple hundred jukebox projects and this was the one that met my needs best. It even lets you play the files ON the file server, if it has a sound card... so my jukebox is a P200 hidden behind the stereo. Webplay can do simultaneous streams to other computers on your LAN too, if you want. Cool stuff.
This is a tuning release (Score:5, Informative)
This is a tuning release. Although all infrastructure like channel coupling is in place, the encoder itself is not ideally tuned yet. One of the goals of this release is to get people to test the new modes and report possible problems (samples were it goofs up). If you do this, be sure to try a blind test. Your mind _will_ play tricks on you otherwise.
Two known problems currently are pre/postecho on some really hard samples, and occasional 'hissing' in the low bitrate modes (< 160).
Both are known and will be fixed in the very near future. RC3 is already expected next week.
--
GCP
Re:iTunes (Score:3, Informative)
See http://developer.apple.com/quicktime [apple.com] for details. Indeed Apple even has a program where they'll put you on their updates system and as an at-need component download. With that in place simply sending someone an Ogg Vorbis-encoded file would trigger their getting the codec automagically.
Re:I give, how does one specify channel coupling? (Score:3, Informative)
Monty
Re:Not all it's cracked up to be... (Score:2, Informative)
One problem I've had with MP3 is that I have yet to find any player that will play consecutive MP3 files with absolutely no break in the audio stream between files. It seems that every one of them has to close the old file, open the new file, read some information, then finally start decoding. In the meantime, there's been a split second break in audio output. Not good for live CDs or any time two tracks continuously merge together. I looked at the API for decoding Ogg to raw audio and it looked perfect to write my own simple player to solve this. I could simply buffer enough audio data that there would be no "skip" in output when switching input files. Perfect.
Then I did some comparisons with sound quality. With Ogg RC1, I encoded part of Mussorgsky's Pictures At An Exhibition to both Ogg and MP3 at around 200kbs VBR. The MP3 sounded perfect but the Ogg had audible clicks and pops. Sorry, but that just wasn't acceptable. So I scrapped that idea and went back to MP3, continuing what I had started. I would have to look into another solution to solve the break-between-tracks problem.
I'm now about 2/3 done with the encoding and this happens? ;-) Honestly I'm not sure whether I want the quality to be improved... If it is, I'll be tempted to start over, which is a lot of work. If it isn't, I don't get the benefits of Ogg... Hmmm. I'll give it a try and see what happens, though.
Re:Ogg Vorbis Quality (Score:5, Informative)
You are comaring:
a) OGG -- decoded stream -- soundcard -- casette input -- amp -- speakers
b) CD -- decoded stream -- CD-D/A-converter -- CD-input -- amp -- speakers
If alternative a doesn't sound as good as b, this doesn't say anything about the ogg-encoding, because it isn't the only variable. Maybe the difference is caused by the different audio-characteristics of soundcard and CD-D/A-converter.
To get a valid comparison, rip the content of the CD as WAV. Then compare the WAV and the OGG, using the same soundcard and the same amp-input. Everything else is totally meaningless.
But even with this setting, there remains one additional variable: your psyche. If ogg and wav were bit-per-bit equal you will still recognize a difference when you know which one of the two you are hearing. So if you want to get meaningful results, you have to make a double-blind-test.
It's really sad how easy it is for the marketing guys to convince people that alternative codecs are inferior, because 95% don't understand anything about scientific methods or statistics. And they will do that, because they have the budget and we have not.
Re:Ogg Vorbis Quality (Score:1, Informative)
You're a little confused (Score:3, Informative)
...Or shall I just stop improving things at this point?
Monty
Finally got through... (Score:2, Informative)
Re:MP3 patent requirements? (Score:4, Informative)
>decoders.
This is wrong. Go take a look at http://www.mp3licensing.com/royalty/index.html
It's not because you don't pay that there are no licenses.
BTW. That page should be _the_ reason why Vorbis will succeed.
--
GCP
Also available at xiph.org (Score:2, Informative)
www.xiph.org [xiph.org]
Re:Not all it's cracked up to be... (Score:3, Informative)
The 128kbps mode is not ideally tuned (IMHO), but the problems are known. 160kbps is already a lot better.
--
GCP
Re:What about... (Score:3, Informative)
So basically this is already done.
--
GCP
Not all it's cracked up to be... (Score:3, Informative)
I compared an Ogg (uning the encoder that came out today) file with a 128KB/s mp3 and a medium quality VBR mp3 (both made with Lame) and I just didn't think the Ogg file was quite there. I was using the same file for all of the tests (Mahler's 9th Symphony).
Also, on my Athlon 900 Oggenc went at 0.6x encoding speed. I usually get 5-8x with 128KB/s mp3s.
I ended up settling on going a little bigger and using a 192KB/s MP3, which I'd say is still the best option.
Best wishes to the Ogg Vorbis team. I hope that I can eventually ditch my mp3s.
Re:Why? (Score:5, Informative)
So, your question is a little silly. The whole point to having multiple bitrates is to allow the user to choose their preferred mix of quality and size. If an aggressive channel-coupling mode can provide significant size savings while having a minimal impact on quality, lots of people will find that worthwhile. And, as you would know if you read the link before posting, you don't have to sacrifice quality at all! Do you really need to ask why someone might want to encode in a mode which has a lower bitrate and identical quality?
Oh, I forgot useful information (Score:3, Informative)
up to 96kbps lossless stereo to 4/6kHz (depending on block), point/6phase/lossless stereo above that
128 lossless stereo to 4/6kHz, 6phase/lossless to 11kHz, point/6phase/lossless stereo above
160 lossless stereo to 4/6kHz, 8phase/lossless to 11kHz, point/8phase/6phase/lossless above
192 lossless stereo to 4/6kHz, 8 phase/lossless above
256+ lossless coupling
The trigger between different stereo models is currently amplitude based (how far the spectral energy is from the noise floor). The trigger threshholds are also increasingly conservative as bitrate increases.
Monty
Ogg Vorbis Quality (Score:4, Informative)
Re:Quicktime component (Score:4, Informative)
Nick D'Amato strikes again- turns out OggDrop _is_ his doing. It does need CarbonLib but seems to not explode when operated, even so.
This is in contrast to the commercial product N2MP3, which faw down go SPLAT! Complete failure in proprietary land. Go get a newer computer, kid. (Yeah right...)
And as a result I have, for the first time, encoded several files into Ogg Vorbis and been able to play them back on a proper set of mastering-ready speakers in an acoustically suitable room... sorta. So, here's my observations so far- some good and some 'whoa! what the heck is that?'.
First of all- Ogg Vorbis DOES NOT lack bass. Trust me on this one. I have some projects being worked on that used my GPLed mastering software to fill in extremely low bass, and I used that to audition Ogg Vorbis. The highs are pleasingly uncolored, a bit 'whiter' than the original recording but it actually seems to help synthetic cymbals. Bright recordings lose absolutely nothing, it's quite impressive really. The lows go down forever, I'm speculating that people are used to some type of midbass wooliness that you get with mp3s? I flat guarantee that the _extreme_ lows get through uninhibited. Almost to a fault...
Here's the joker in the deck: every tune I encoded and played back was somewhat choppy- and VARI-SPEEDED.
I can't begin to imagine what would be causing that. It's really being done pretty damn well! It sounds very, very much like the original recording, with a bit of interference and choppiness, except the tempo is _significantly_ slower. Like more than 10 bpm slower. This is a pretty serious problem... and I don't believe it can be part of Quicktime because Quicktime has been able to avoid that sort of thing for many, many years.
A 300mhz G3 machine with 128M of RAM ought to be able to deal with this- if Ogg Vorbis is truly that processor-destructive that's a serious objection to it. It'll never work in embedded apps or portable players if it has to eat that much CPU. I'm hoping it's a bug. Actually I _know_ it is a bug, because dropouts are one thing with inadequate (ha!) CPU, but _varispeed_ should not be happening. There's no excuse for that as a reaction to inadequate CPU.
So, all told, I am delighted with what I've learned. And even with the problems I encountered, I can confirm that Ogg Vorbis _does_ have bass, deep bass, and that its tonal character, even at 128K, is quite impressive. If I was mastering for it I'd master stuff for soundstage depth knowing it would drag all the highs and lows out that it could, that it would make things 'whiter' and zippier kind of like Fraunhofer MP3 encoders, but in a less intrusive yet more effective way.
At the same time, this port of it is still 'freaky bizarre demo' quality and could not be used professionally. I'd love to know if this varispeed is happening strictly on playback- that would mean I had a commercial-quality free Nick D. _encoder_ and just didn't have a playback mechanism that worked properly.
Believe me, guys, I'm rooting for you. But I can't do this work for you, because I'm not a C-slinging programmer gunslinger. If I was, I'd have been trying to help out loooooong ago. Best I can do for now is state unequivocally: yes, Ogg sounds better than MP3 if you like clarity and wide-range frequency response. I look forward to when it grows up and can support platforms such as mine. I can't really give it a full-on audition, or include it in the mp3 study I did, because it's just plain not ready and not working reliably, but finally, at long last, it's working unreliably and that is enough to give me a taste. And I like where it's heading.
-Chris Johnson
Re:Ogg is the great OSS success story... (Score:3, Informative)
Actually, the libraries are BSD-licensed so companies will be more likely to adopt it.
Re:What about... (Score:1, Informative)
While I haven't been able to get RC2 yet... (Score:3, Informative)
Like the topic says, I haven't been able to get to RC2 yet, thanks to it being slashdotted, but I seriously doubt RC2 sounds worse than beta 4, and while encode times _are_ slower than mp3, they're nowhere near as slow as some people are saying. (I get about 3x speed on my Duron 850 with b4). Clicks and pops are likely a cause of a bad rip from the CD, not the encoder.
I've been using nothing but Ogg for my CDs for a while now, and have encouraged many friends to do the same. People really need to give Ogg a fair, unbiased try before they go saying it sucks, because it's most definitely at the very least, better than mp3 at the same bitrate. Check out PCABX [pcabx.com] for info on how to do a good double-blind listening test.
Congrats to Monty and the rest of the Ogg Vorbis team. Keep up the good work.
Re:Dolby Digital (Score:5, Informative)
I don't think there are any un-patented 5.1 channel codecs around.
Actually, if you look at the last answer on the Vorbis FAQ [xiph.org] you'll see that Ogg Vorbis already supports encoding of up to 255 channels per stream, so, theoretically at least, it ought to be a cinch to use Vorbis for 5.1 audio.
This could be a real opportunity for Ogg to become the first mainstream audio codec to support 5.1 explicitly. It would be a real leg-up for Ogg's chances if it gets accepted as the choice of audiophiles, and having 5.1 supported before MP3 and WMA can only help with that. Those who have experimented with DVD Audio would finally have a format worth considering for ripping purposes, and it helps that Vorbis sounds very musical.
Once and for all: Reliable testing for quality (Score:4, Informative)
Encode MP3s, RC2-OGGs, and whatever else you like, at all the bitrates you are interested in. I recommend doing this for many different types of music you like.
IMPORTANT STEP 1:
Once they're on your computer, decompress them back into a .WAV file. Make sure you keep track of which .wav came from which compressed file. If you tested both MP3 and OGG at 3 different bitrates each, you will have 6 .WAV files for each song, plus the original .WAV (don't delete it). Then cut out relevant passages from each of the songs, maybe a minute each, with a wav editor.
IMPORTANT STEP 2:
Once you have these wav files on your hard drive, tell your roommate to burn them on a CD, in an order that he will write down but not reveal to you. Then put the CD into your stereo and get a good paid of headphones. Crank it up, and take notes on which versions of the passages sound the best and why. See how successful you are in identifying the original wav file when you don't know which it is. See if there is any pattern to your responses.
Until you do a double-blind test like this (come on, it's not difficult) you really shouldn't be shooting your mouth off about which format sounds better.
Re:Not all it's cracked up to be... (Score:3, Informative)
1) I'm doing encoding on Mahler's Symphony No. 9 because I'd like to have a copy at work too (without leaving my CD up here). The idea is that even though I'm using lossy compression to not be able to notice it. Also, classical is much more demanding on an encoder so I thought it would be a better test. I also feel compelled to point out that CDs are a lossy format. Heck, why record anything, you're always losing data.
2) I'm listening through pretty high end headphones, Sony MDR-V600 ($120 at Best Buy), so sometimes I can notice things that others can't.
3) Yes, I'm sure I'm using the current version of Ogg. I'm on the devel mailing list (I'm planning on adding Ogg support to the MP3 Tagging software that I wrote, QTagger.) and saw the annoucement come out and installed the RPMs this morning. I upgraded from Beta 4 which came with Redhat 7.1.
So what are the differences?
*) The sound on Ogg files sounds clunky as it changes bitrates. This was especially noticeable on the recording I was compressing since it was originally analog and had a constant (though slight) background hiss. The noticeable changes in what should be a constant sound were quite distracting.
*)To Ogg's credit, they don't have as noticeably the fluttery sound of compression artifacts that you sometimes notice in MP3s. Lame is a nice encoder though, so with the -h switch these normally aren't too bad. I don't hear them at all in 192 KB/s MP3s, which I reencoded all of my classical in today. I use 128KB/s for rock and jazz. It would be interesting to go back and repeat my test with something idiomatic from those genres.
*)I thought Lame's VBR did sounded better than Ogg Vorbis and they seem to be similar schemes. There was a noticeable squeaking sound in the background on the MP3, but it was still clearer than the Ogg file.
For summary, since I'm listening on pretty hi-fi stuff, I can hear all of the little background-ish type things. I prefer these to be constant as opposed to variable. If there's analog hiss, it should sound uniform across the recording.
I'll repeat, I'm glad Ogg's around and I hope it improves, but I'm just not ready to switch yet.
Re:I give, how does one specify channel coupling? (Score:2, Informative)
you can't yet (it's not a simple switch, it affects the entropy model to use). I *will* make user-specification of stereo model possible as soon as possible.
Monty