Ikonboard » Printable Version of Topic

Forum: Geek Forum
Topic: Ogg Vorbis VS MP3
started by: Blain

Posted by Blain on Sep. 10 2001,01:58

A little while ago cr0bar recommended that Catknight ditch all his mp3s’ and rip his music as an Ogg Vorbis format instead, because it sounded much better. The possibility of a noticeable increase in music quality greatly intrigued me (I have a VERY nice sound system) and when I found this thread I immediately downloaded an Ogg encoder. I proceeded to rip a song (Shastakovich’s fifth symphony, first movement) at 320 kbps (what I use for a standard) in .ogg, .mp3, and .wav and listen to them. The .wav and .mp3 sounded very similar (nearly indistinguishable), but the .ogg sounded very noticeably higher in pitch, while about the same in clarity. I also noticed that the .ogg file was 16:16 minutes long, whereas the .wav and .mp3 were both 17:42 min long, this leads me to think that the .ogg is simply sped up in order to achieve better treble.

I only downloaded a few .ogg encoders (for this experiment i used LitexMedia) and they all either limited my encoding rate or had a limited trial period, so if anyone (cr0?) could suggest a GOOD free encoder that I could try, I would be very appreciative. I am all for a better, more efficient method of encoding music; but, so far, Ogg Vorbis hasn’t achieved that.

Edit: I am winning a game of “drink until you fall down” do you really think I can post a coherent message?

This message has been edited by Blain on September 10, 2001 at 09:03 PM

Posted by CatKnight on Sep. 10 2001,02:37

wow blain I was about to post this topic. Anyway I have done my own tests and found that ogg 192, lame --r3mix, and the original cd wav, are all indistinguishable. however, the r3mix clip is considerably smaller then the ogg 192. don't believe me? < test it yourself >.

Posted by Blain on Sep. 10 2001,02:54

This might just be my drunken ass talking, but did you send me to an audio testing site (i didn't spend a lot of time there)?
I was talking more about my personal experience, and I can most definitely tell the difference between a 192 Kbs mp3 (I believe lame), ogg, and wav ( I am unfamiliar with r3mix); which is why I encode at 320 Kbs. Even at 320 Kbs I can still easily tell the difference between listening to a CD on my computer (digital out) and an mp3 on my computer, that’s why I wondered if I possibly downloaded a shitty encoder when the mp3 sounded better then the ogg.

This message has been edited by Blain on September 10, 2001 at 09:58 PM

Posted by ASCIIMan on Sep. 10 2001,03:53

Okay, get < Exact Audio Copy >. Best ripper / encoder interface / audio burner hands down.

I rip with lame --r3mix. I find it pretty much indistinguishable from the original .wav (except for some things like white noise, extreme percussion, etc., but pretty much the only way to fix that is to use < lossless compression >). Except for those isolated events I mentioned above, I can't tell the difference using my super-duper headphones and 1337 listening skillz. < More on the --r3mix setting. >

edit - Oh yeah, EAC can use pretty much anything to encode. It has built-in support for most of the mp3 encoders, ogg, and the popular lossless formats, and will support anything with a command line interface.

And, Blain, I seriously doubt you can tell the differance between 320 and the cd, unless you're running at CBR. VBR can actually have higher quality than a CBR 320 kbps mp3.

This message has been edited by ASCIIMan on September 10, 2001 at 10:58 PM

Posted by CatKnight on Sep. 14 2001,13:53

yeah blain is full of shit. he thinks he can tell the difference between two cd quality samples. he won't even try to test them with a double-blind wav compare program.

Posted by RenegadeSnark on Sep. 14 2001,15:35

Ogg, while it should sound excellent at 320 kbit like all the other formats, beats MP3 256Kbit at its own 192kbit VBR. MP3's sound "clipped" - that is, the top and bottom of the sine wave of sound are normalized. This also tends to flatten out the spatial and density qualities of many songs, which Ogg does not do.

As for your songs being different lengths, you're way off base. I don't know how you ended up with a shorter file, but I'd look into your encoder.

Personally, my rips are done with ripper-X which shells to the GNU/Linux ogg encoder and CD Paranoia as a ripper (set at 1x digital, maximum paranoia). I get really fucking fantastic audio out of that.

Posted by CatKnight on Sep. 14 2001,20:46

hey snark I'm talking about comparing a lame --r3mix mp3 to a 192 vorbis one. the r3mix ones usually average around 170-190 kbps and sound completely transparent. I'm not bashing vorbis, I think it sounds fine. I'm just saying that there is no advantage in using vorbis since the files aren't really any smaller, and ogg files are not supported by as much software/hardware, and are not very popular.

Posted by RenegadeSnark on Sep. 15 2001,01:03

A stock Ogg versus a --r3mix. I can actually tell a difference... many people can, and I'm not bashing your ears or anything but even --r3mix seems to have artifacts.

Besides, I use Ogg because I like to be able to support a patent-free, open-source format. That's also one of the many reasons I think PNG is better than JPEG (that, and the fact that PNG images look much better and support 48 bit color with a full channel of alpha transparency).

Viva la revolution!

Posted by CatKnight on Sep. 15 2001,02:00

I'm going to have to ask you to get that audio compare program from the link I mentioned above, make 3 sample wav files (1 cd rip, 1 r3mix=>wav, one ogg=>wav), and run all three. I seriously doubt you will be able to hear a difference, unless you have a บ,000 stereo system or electrostatic headphones hooked up to your puter. I thought I would be able to, but then I tried it and I wasn't. Use normal clips btw, not some strange random clips that just happen to show a minor difference.

Posted by RenegadeSnark on Sep. 15 2001,02:09

My father and I have really good ears... in the living room he spent บ,000 on his system - a Paradigm 5.1 set of speakers and a Marantz SR-19 receiver... the backyard speakers alone are a 񘈨 set of B&W's.

I have a Cambridge Soundworks DTT-2500. It sounds pretty good, and I can tell a difference between Ogg and mp3. Ogg sounds closer to WAV.

Posted by miNus on Sep. 15 2001,02:12

quote:
Originally posted by Blain:
Edit: I am winning a game of “drink until you fall down” do you really think I can post a coherent message?

There. Does that clarify some things for you guys?

Posted by ASCIIMan on Sep. 15 2001,03:08

quote:
Originally posted by RenegadeSnark:
I have a Cambridge Soundworks DTT-2500. It sounds pretty good [...]

Ummm... yeah I have a set of those too. They sound like crap. Your opinion on anything relating to audio quality has now been invalidated.

Posted by CatKnight on Sep. 15 2001,04:15

I agree. I did my sound tests in my room with everything turned off so it was silent. I used my sony v-6 headphones, which are what most radio dj's use. Snark, until you do this double-blind test, you are just kidding yourself.

here I made it easy for you. Here is a direct link to the program: < http://www.personal.psu.edu/dbl125/ABX173_setup.exe >

it's pretty self explanatory. just plug in the cd wav to B and the sample wav to A, and run the test. press A or B to listen to each sample, then X to hear a random one. click which sample you think it is, and then click next trial. do about 10-20 trials and see your score. if you get less then 70\% then it's likely that you were guessing.

one more note: don't test mp3 vs ogg, because it's likely you will hear a difference since they are both lossy compression. run mp3 vs. cd and ogg vs. cd. you probably won't be able to tell the difference between any of them, which is my whole point. ogg doesn't sound any better then a properly encoded mp3, and is less supported. then again if you like open source software then knock yourself out. I'm just trying to prove that mp3 is not of any lesser sound quality (but happens to be smaller too)

last last note: by properly encoded mp3 I mean --r3mix with lame (preferably 3.90).

This message has been edited by CatKnight on September 15, 2001 at 11:29 PM

Posted by CatKnight on Sep. 15 2001,04:32

oh I just noted your comment about mp3's sounding clipped. that is due to 100\% dB peak level on the cd. the --r3mix switch gets rid of this distortion by brining the total sound level down to 98\% using a linear method. I suspect ogg does the same, thats why it sounds better to you then a mp3 encoded with less then optimal settings.

Posted by RenegadeSnark on Sep. 15 2001,06:11

quote:
Originally posted by ASCIIMan:
Ummm... yeah I have a set of those too. They sound like crap. Your opinion on anything relating to audio quality has now been invalidated.

Are you sure you didn't accidentally try to shove the speakers up your ass before listening to them? I'm not going to claim that the DTT-2500 set is as good as a movie theatre's systems, but as far as PC speakers run the DTT-2500 is a good, solid speaker set.

Posted by RenegadeSnark on Sep. 15 2001,06:17

Don't forget that with .ogg files you can have more than two channels - this means that 5.1 audio encodings are going to become a reality.

Posted by RenegadeSnark on Sep. 15 2001,06:23

quote:
Originally posted by CatKnight:
oh I just noted your comment about mp3's sounding clipped. that is due to 100\% dB peak level on the cd. the --r3mix switch gets rid of this distortion by brining the total sound level down to 98\% using a linear method. I suspect ogg does the same, thats why it sounds better to you then a mp3 encoded with less then optimal settings.

I'm going to bite on this and try your test. But I personally find it remarkable that an audio format created by a group of open source developers using non-patented technology in its *default settings* can beat a frauenhoffer mp3 with default settings, and rival if not beat the highly customized encodings of a specific encoder.

That, and I'm making an effort to give Frauenhoffer as little support as possible, since they are now trying to charge royalties on open source developers. Don't be suprised when the lame becomes a piece of history because Frauenhoffer decides to excersize their patents. When that happens, ya'll are welcome over here in the Ogg camp. We're having a hell of a party.

Posted by ASCIIMan on Sep. 15 2001,06:41

quote:
Originally posted by RenegadeSnark:
Are you sure you didn't accidentally try to shove the speakers up your ass before listening to them? I'm not going to claim that the DTT-2500 set is as good as a movie theatre's systems, but as far as PC speakers run the DTT-2500 is a good, solid speaker set.

Okay, the UBB gnomes ate my post, so I'll make it short:

DTT2500 crossover: 155Hz
Meaning: Those 2.5" drivers have to reproduce everything from 155Hz to 20000Hz.
Result: Crappy Total Harmonic Distortion and Linear Distortion figures, entire frequency bands are lost.

Total Harmonic Distortion, Distortion (not Signal-To-Noise: Creative/Cambridge don't list any specs
Meaning: It's not listed because it's crap (most likely >10\%). They know it's crap so they don't list it in their specs.
Result: The average consumer won't know to ask about it and buys a 赨-300 set of speakers that sound of crap ("Ooooh... but it's got 5.1 Dolby Digital surrond... Oooooh." Yeah, right ).

Try running a frequency sweep on them sometime. Notice how the sound seems to fade in an out on different portions? That's what it's doing to what you're listening to.

Changing the subject a bit, anyone want to buy a set of DTT2500's cheap? They're great speakers, I swear!

BTW... If you wanna get a better pair of speakers for not a whole lot more money, get some of < these guys >.

This message has been edited by ASCIIMan on September 16, 2001 at 01:56 AM

Posted by RenegadeSnark on Sep. 15 2001,07:11

Klipsch does make excellent speaker systems. I actually got the DTT2500 set as a christmas present a few years ago. Yeah, so the crossover is actually about 140Hz. Klipsch doesn't always make an effort to advertise their THD either.

Audio is a really interesting thing. As it is, *all* methods of recording audio (LPs to casette to CD to MP3) rely on taking many samples per second, like video relies on taking many "pictures" per second. Audio is really a vibration in the air, and the only reason we can "hear" it is because our ears are designed to pick up this vibration and interpret it in a certain way in our brain.

Did I ever once claim that the DTT2500 was the end-all best sound solution? Hell no, there are much better sets out there. But there are also much crappier sets (Bose included) that lurk around. Generally what I've found is that if you can buy it at Wal-Mart, Radio Shack, or Fry's Electronics, it's not a good speaker set.

You can have your Klipsch, and when I want to do some *serious* listening, I'll sit down in our living room to listen on my dad's 5.1 set of Paradigm speakers. Some of their low end equip has a THD of less than like 0.3\%.

Oh and CatKnight - I have a set of MDR-V6's too. I love them.

Posted by ASCIIMan on Sep. 15 2001,07:46

quote:
Originally posted by RenegadeSnark:
Klipsch doesn't always make an effort to advertise their THD either.

True, but in general (and for these systems, both of which I've listened to (although I only listened to the Klipschs after I had bought the DTT2500s)), tweeter + mid + sub systems have better sound than mid + sub systems.

quote:
Originally posted by RenegadeSnark:
Did I ever once claim that the DTT2500 was the end-all best sound solution? Hell no, there are much better sets out there. But there are also much crappier sets (Bose included) that lurk around.

True, but my point was that they don't sound great, especially when you can buy better sounding systems for about the same price.

quote:
Originally posted by RenegadeSnark:
Generally what I've found is that if you can buy it at Wal-Mart, Radio Shack, or Fry's Electronics, it's not a good speaker set.

You can now buy both the Klipsch and Cambridge speakers at Best Buy and CompUSA.

quote:
Originally posted by RenegadeSnark:
You can have your Klipsch, and when I want to do some *serious* listening, I'll sit down in our living room to listen on my dad's 5.1 set of Paradigm speakers. Some of their low end equip has a THD of less than like 0.3\%.

Umm, yeah... You do that. I have a 7.1 system with some Linn speakers and NAD stereo components, but I'm talking about compact that doesn't require a whole rack so I can use it easily on my computer audio here.

And please don't take my previous posts (or this one) as a flame, they weren't meant that way.

Posted by kornalldaway on Sep. 15 2001,11:50

i got a question to all those who actually know what they are talking about....
with kame --r3mix you can set output sampling frequency to 48kHz rather then default 44.1kHz. Does the 48kHz have any major advantages over the default 44kHz in quality of sound and what about file size?

Posted by ASCIIMan on Sep. 15 2001,15:37

If you're encoding from a CD, it's better to encode at 44.1KHz because that's what sampling frequency the CD is at. Encoding at 48KHz would force a resampling which would introduce additional artifacts as well as causing the mp3 to be of lower quality because more samples/second + same bitrate means the encoder has to work harder to encode sounds that wouldn't have been in the original 44.1KHz audio track.

Posted by CatKnight on Sep. 15 2001,18:04

quote:
Originally posted by RenegadeSnark:
I'm going to bite on this and try your test. But I personally find it remarkable that an audio format created by a group of open source developers using non-patented technology in its *default settings* can beat a frauenhoffer mp3 with default settings, and rival if not beat the highly customized encodings of a specific encoder.
That, and I'm making an effort to give Frauenhoffer as little support as possible, since they are now trying to charge royalties on open source developers. Don't be suprised when the lame becomes a piece of history because Frauenhoffer decides to excersize their patents. When that happens, ya'll are welcome over here in the Ogg camp. We're having a hell of a party.

Yes, I agree. Vorbis is a great encoding algorithm, and it is open-source too! If mp3 ever crap out and starts charging money or require songs to be registered or something, then I'll fall back to ogg.