Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: MPC vs. ogg vorbis? (Read 11340 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

MPC vs. ogg vorbis?

Ok folks, I can understand if you get mad now, because you've already read millions of these posts, but please excuse my lack of wisdom...


I've read many reviews and tests already, but some of them said ogg is better, others glorified mpc and I got more and more confused.  ???

Since Ogg 1.0 is released now and they said, it provides some improvements also in the high-bitrate area, I'm now interested how it compares to MPC.

I'm interested in bitrates around 160 - 200 bps. (ogg: -qual 5).  How does it compare to MPC on similar rates?

And please don't tell me just "ogg is better" or  "mpc rulez!", I'd be very glad if you could explain why you think the encoder you prefer is better.

Thank you very much for your help, it is highly appreciated!!

MPC vs. ogg vorbis?

Reply #1 –
Ogg is better.

HTH.

MPC vs. ogg vorbis?

Reply #2 –
Quote
I'm interested in bitrates around 160 - 200 bps. (ogg: -qual 5).  How does it compare to MPC on similar rates?

At these bitrates mpc is unbeatable and provides the quality no other lossy codec can achieve. This *may* change in the future, but for now musepack 0wnz all. :-)

MPC vs. ogg vorbis?

Reply #3 –
Hmm, thanx very much for your opinions. It's just like I've expected it, till now 50% prefer ogg and 50% MPC

Ok, could you please tell me the site for the latest MPC releases, the one for OGG I know ( http://ogg.benjamin-lebsanft.de/ )

And I hope we'll get some more views here. So thank you again in advance!!

MPC vs. ogg vorbis?

Reply #4 –
Quote
Originally posted by eloj
Ogg is better.


This, of course, depends on how you define "better". Since Xenoide seemed to have his focus on sound quality at 160 - 200 kbps, Ogg Vorbis is not better. At >160 kbps, MPC outperforms Ogg Vorbis.

MPC vs. ogg vorbis?

Reply #5 –
On pre-echo testing (with castanet samples), it's easy for me to ABX Vorbis I up to 400 kb/s (At 400 kb/s, 25/30, at -q9, 29 or 30/30 kb/s]. With Musepack, I can do it with standard, extreme (more difficult), but not with the new insane profile...
At -q5, Vorbis is simply horrible !!!
It's only pre-echo, sure - and Vorbis did a good job on normal music, better than mp3. But I can't find a more universal sample sample than castanets.wav, and another sample I can ABX with --extreme on mpc.

I did my choice...
Wavpack Hybrid -c4hx6

MPC vs. ogg vorbis?

Reply #7 –
Hmm... Since the same questions seem to be asked around here quite often, we really need to compile a master FAQ that has all the answers to the "this vs. that" questions and the "i have the best preset" questions, etc...

Dibrom?

MPC vs. ogg vorbis?

Reply #8 –
Quote
Originally posted by Xenoide
Ok, could you please tell me the site for the latest MPC releases

I would recommend my site, http://www.saunalahti.fi/~cse/mpc/.

MPC vs. ogg vorbis?

Reply #9 –
Quote
Originally posted by guruboolez
With Musepack, I can do it with standard, extreme (more difficult), but not with the new insane profile...


Wow! Good hearing... What ABX results did you get on those? In which way was MPC distinguishable from original? Pre-echo?

MPC vs. ogg vorbis?

Reply #10 –
If anyone is interested for testing this damn artifact (I'm one of the first to be pre-echo sensible , in fact for example every non-LAME MP3 really sucks, what "chewingum" effect! argh... ) I found some pre-echo sensible content in several Gipsy Kings songs,
where guitars rhythms cover high contrast (sharp) peaks, well distinct (so any defect is more easy to ear), plus by percussion ... and from the fact that music is "vibrant" itself

P.S.: hope that Vorbis will get even better in the future to avoid this artifact. Even if must mention that now handles already good and better the MP3 (Lame included).
"Taking a jazz approach and concentrating on live playing, I wanted to use several different rhythm sections and vintage instruments and amps to create a timeless sound that's geared more around musicality and vibe than sonic perfection. The key was to write with specific rhythm sections in mind, yet leave open spaces for soloing." Lee Ritenour

MPC vs. ogg vorbis?

Reply #11 –
I was really surprised too 

The problem was only pre-echo, and i can hear it with a crappy headphone as well with a good model (2€ vs 250 €).
The tests samples are castanets.wac, and an another, submitted by Garf if I remember, castanets too, but more precise, and more easy to identify....

With this last sample, I can ABX 30/30 PsyTEL 2.15 « Ultra » : I was very disappointed by this score. But CBR 512 did a very good job (it isn't CBR : strange...)
Vorbis was very easy up to -q 8.5 (and 8 with traditional castanets.wav), a little more difficult at -q9 (something like 27/30), harder at -q 9.5 (23-25/30 but with a lot of concentration), impossible with -q 10.
Musepack have a strange sound : not really pre-echo, but a brighter or metallic coloration : I can explain it, even in french. It is a very little « piiou-piiou » sound (pre-echo make tchac-tchac, and not tac-tac).
With mpc 5 (standard) it is now very easy for me to hear it. But don't forget that I hear castanets.wav thousand times, and with a new and similar sample, I have difficulties to hear a problem (but not with Vorbis ou PsyTEL : pre-echo is pre-echo). Mpc6 don't correct the problem : ABX gives me something like 24/30, sometimes more, sometimes less. Beyond the score, with ABX, there is the certitude [does this word exist in english ? Sorry if not...], the evident feeling at moment to hear a real problem (at 5 or 6 times on a 30 test) : and the results gives me ALWAYS reason. I had the feeling to heard something strange on the encoded file, and it was always true.
Mpc7 is not really significant (20/30, and sometime less). But this strange & certain feeling (at 4 or 5 moment) was always good (5 time on 6, something like this)
Mpc8 is only luck (10/30 to 18/30)... I sometimes have the same certitude (5,6, 8 times on a test), but the result show me that it wasn't always exact : I was certain to hear a real problem on original file  No more problem with --braindead...

Musepack have maybe others problems : but I never heard it. But I perceived [does this word exist too ? ] with certitude artifacts on Vorbis and PsyTEL at the highest bitrate on an extreme but universal sample. For me, thats enough to not trust theses codecs for archiving or a very very high listening pleasure. Thanks to Mr Buschmann, Klemm, Piecha & another...
And sorry, again and again, for english
Wavpack Hybrid -c4hx6

MPC vs. ogg vorbis?

Reply #12 –
Your impressions about MPC coloring the sound on this sample coincide with some other people's impressions. What encoder version did you use?

20/30 is below 5% probability of guessing, it's pretty probable that you were actually  hearing a difference at mpc -q7.

Nice to have here a person with such good hearing.

MPC vs. ogg vorbis?

Reply #13 –
Man you have gracious ears, i can't tell the difference between vorbis and AAC at -q7 (somewhat 224kb/s). The only thing im sure is that they sound different, but they both sound good enough to archive IMO, on the other hand -q6 isn't enough to archive in both, i can easily tell the difference, but i'm useless -q7 and above.

So based on your experience, tests and preferences, which one do you think is beter at -q7 or -extreme... Psytel AAC Encoder 2.15 or Vorbis 1.0???

By the way, -q9 or ultra are just useless IMO, i would rather use a lossless codec instead like monkey audio...

MPC vs. ogg vorbis?

Reply #14 –
hehe, pretty amazing indid, there you have one person for your high-bitrates test mr. ff123 

'One Ear to bring them all and in the darkness bind them'

no, seriously, while i did some lame tests some time ago i found a person with similar capabilities, i didnt do a lot of tests to be a real beliver, but maybe now its the time to repair that mistake, since i will start to belive that there really are ppl with such amazing ears.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

MPC vs. ogg vorbis?

Reply #15 –
Quote
Originally posted by KikeG
Your impressions about MPC coloring the sound on this sample coincide with some other people's impressions. What encoder version did you use?

20/30 is below 5% probability of guessing, it's pretty probable that you were actually  hearing a difference at mpc -q7.

Nice to have here a person with such good hearing.


My hearing isn't so good. Last year, I was in trust with Fraunhofer/Radium at... 128 kb/s  I just repeat, I was the first to be flabbergasted by these results (I repeat the tests day after day, during a whole week, just to believe it !).

The encoder was mppenc 1.06 (I hadn't internet in july [only 1 hour connexion for the entire month], and I just learn today that an updated version were released by Frank Klemm...). Decoder : 1.06 too...
I tried older versions, especially Buschmann 1.79 one : no amelioration, and no degradation (or very subtile : but if I can hear a problem, evaluated it is another very difficult thing)

Insane is not so concluant. I can all days ABX castanet on these profile. And extreme too need a little « échauffement » [I don't know the word... my hear must be warm, trained like an athlet before his jump], but it's very fast too hear a problem with extreme. Insane, I don't know why, is less concluant... maybe concentration, ar auditive memory...


I tested the codec on a crappy headphone. It is funny, but interesteting. Musepack particular artifact is less evident with the bad hardware sound. But traditional pre-echo was always here (note that I tested Vorbis during this test, but not PsyTEL). I ABX vorbis -q8 easily (19/20), and a god 15/20 with -q8.5. With mpc, I was first in absolute guessing... but the last 10 sessions were more concluant with standard profile (4/10 and a nice 9/10 on the second half). Hard to conclued anything : immediatly, my hearing is not in trouble with mpc ; but with an attentive comparison, a evident default is appearing.
With a sample I never heard, I can surely identify Vorbis, or PsyTEL on pre-echo (if sample is strong enough, of course : not on a string quartet). But I doubt that I can did it with mpc standard... The difference is more than 100 kb/s (170 kb/s vs 300, and up to 400 kb/s for Vorbis). But to be totally in trust, I encode all my CD in --insane.
Wavpack Hybrid -c4hx6

MPC vs. ogg vorbis?

Reply #16 –
Well, when I may summarize it a bit, would it be right to say

MPC offers better quality than ogg 1.0 in the 160-200 bps range?

And to make it clear, I'm talking about usual music, classical or electronic.
I don't really mind if someone can hear some artifacts in very special test samples you won't find in normal music anyways.

(Ok, castanets.wav is perhaps a good example for music)

MPC vs. ogg vorbis?

Reply #17 –
Quote
Originally posted by Phobos
Man you have gracious ears, i can't tell the difference between vorbis and AAC at -q7 (somewhat 224kb/s). The only thing im sure is that they sound different, but they both sound good enough to archive IMO, on the other hand -q6 isn't enough to archive in both, i can easily tell the difference, but i'm useless -q7 and above.

So based on your experience, tests and preferences, which one do you think is beter at -q7 or -extreme... Psytel AAC Encoder 2.15 or Vorbis 1.0???

By the way, -q9 or ultra are just useless IMO, i would rather use a lossless codec instead like monkey audio...



Don't be so elogious. I never believe to achieve these results... but with my dead internet connexion, and a new fantastatic codec (I tried vorbis at -q3 and in winamp, it sound good... so -q7 seems to be an himalayan bitrate !!!), I was obliged to test myself, and not just reading CitAy or garf sessions resutts. Try with ABC/HR (the first 1,5 second of castanets.wav) : its obvious. Try, I insist : it seems impossible, and it was so easy.

I said that PsyTEL disapointed me at ULTRA profile (320 kb/s), the biggest VBR profile. Vorbis did a better jog, but with more than 400 kb/s.
Yesterday, I launched a fight between Vorbis and PsyTEL at medium bitrate (understand 160-250 kb/s  ]. I was very attentive on notation, in order to « hierarchise » these too codec (and mpc too, but I knew his - first - position before testing). The soft was ABC/HR (thanks, ff123 : this software is a fantastic scalpel).
I can publish the whole notation (comments are in french, I don't post it) :

Ogg -q5 => 1.1
Ogg -q6 => 1.6
Ogg -q7 => 2.1
AAC Normal => 2.7
AAC Extreme => 3
Mpc Standard => 4.3

Vorbis -q5 was horrible (in french). -q6 is better, but awfull too. -q7 is better, but too easy and so bad...
AAC were at evidence better : better pre-echo control. Some time I notice a bad noise on the guitar : the same thing we can easily heard with mpc --radio, which use PNS. It is very subtle, of course, not like mpc --radio.
MPC was evident to identify, but it's not annoying.

I did 8 sessions for this test (I'm able to do 100 session, but notation and commentary are so annoying). I « ABX » all the codecs  with 8/8, except mpc (7/8, I was not able to identify it the fist time). The placebo sample (the original) was accidentally attribute to mpc on the first session, and not on the seven others.

More interesting : I always IDENTIFY the codecs family, and often attribute the good filename (i recognise ogg defauts : hard pre-echo, a brighter noise an guitar ×× AAC : less pre-echo, but sometimes a growing and inatural noise on guitare ×× mpc : can really explain, not pre-echo, but another coloration on castanet attack). Vorbis was easy to hierarchise, but AAC not : sometimes PsyTEL normal seems to have better pre-echo control, but a strange coloration on attacks, like mpc, but louder. AAC extreme didn't have these coloration but pre-echo didn't appear me to be better than « normal » profile. I identify correctly aacN and aacX 4 times, I hesitated 2 times, I did an intervertion at 2 times.


For me, it is evident that aac have a better responce to pre-echo than Vorbis. AAC normal (185 kb/s - traditionally 170 I think) is better to that point that Vorbis -q7 (236 kb/s, 224 habitual), and certaily better, or similar to -q8, maybe -q8.5 (not -q9, I'm sure of it).
Wavpack Hybrid -c4hx6

MPC vs. ogg vorbis?

Reply #18 –
man thans a lot, that second point of view is all i needed to be sure to make AAC my all-in-one codec  im so happy and thnx to rjamorim for introducing me to this amazing format. Now i can't wait for main profile

MPC vs. ogg vorbis?

Reply #19 –
Quote
Originally posted by Xenoide
I'm interested in bitrates around 160 - 200 bps. (ogg: -qual 5).  How does it compare to MPC on similar rates?

Both perform very well at this bitrate so you can't say one clobbers the other. However, all things being equal, I would select MPC because it was made from the ground-up to be a higher-bitrate encoder. It's harder to find artifacts with MPC @ 160-200kbps than Ogg Vorbis.

MPC vs. ogg vorbis?

Reply #20 –
well, thanks a lot.

after all I've read now, I expect better results with mpc. So now I'll do some ABX tests and check how much difference I'll hear...

MPC vs. ogg vorbis?

Reply #21 –
What's intented for coloration?
An extraneous signal mixed to the original, or like color/chroma smearing in image compression?

excuse me.
"Taking a jazz approach and concentrating on live playing, I wanted to use several different rhythm sections and vintage instruments and amps to create a timeless sound that's geared more around musicality and vibe than sonic perfection. The key was to write with specific rhythm sections in mind, yet leave open spaces for soloing." Lee Ritenour

MPC vs. ogg vorbis?

Reply #22 –
Quote
Originally posted by unplugged
What's intented for coloration?
An extraneous signal mixed to the original, or like color/chroma smearing in image compression?

excuse me.


I can't really explain it. I hear perfectly the artefact now on castanet : the attacks remain strong ; the sound is sharp (others codecs are smearing the castanets attacks : it's why it is really easy to hear with a poor headphone - Dibrom often repeated it).
But there is something else, not near the attack, but IN the « Tac - tac - tac ». Difficult to explain.

Have you ever thrown a stone on a frozon « lake » (smaller than a lake, of course...) ? Just throw a stone high in the sky : when it fall on the ice, you can hear a sharp but little distorded sound, like a brief « piouiii ». Musepack artifact sounds someting like this : not a metallic coloration but an icy sound  (I remember, that some people find mpc cooler than mp3  That's not entirely false  ). Perceptible if you know the original coloration of the instrument/sample, not really annoying to my taste. But absolutely obvious : it's not transparent on this sample.

But what I hear with the natural sound of a castanet disapear with an electronic attacks. On Frank Klemm's page, there is a funny sample, called short_block_test : one minute of wave file, but only 5 ko on a rar file  Musepack is encoding it very well with standard. I can not identify ONE artifact on the whole sample (ABC/HR is necessary if you don't want to be def). Bitrate explode : more than 300 kb/s (with a jump at 600 kb/s on the end).
With Vorbis, it's really funny. I tried to encode it with -q10 (libvorbis 07/13/02 with oggdropXP) : 140 kb/s, and not really 500... No problem, there is a CBR setting : 96 kb/s with ogg CBR 500 ??? Very strange. Quality is bad : very very easy to ABX with -q7, very easy with -9, simply easy on -q10... AAC had less pre-echo problem if I remember correctly, but more bitrate :


mpc5 :  335 kb/s [2776 ko]
ogg7 :  101 kb/s [838 ko]
ogg9 :  132 kb/s [1098 ko]
og10 :  140 kb/s [1161 ko]
aacU :  170 kb/s [1410 ko]
aac512 : 250 kb/s [2075 ko]

aac CBR 512 is only 250 kb/s... Ivan, what happened ? It's the only file, with mpc5, I couldn't identify !!!


Strange and different VBR behaviour...
Musepack is more adaptative : bitrate explode, but quality too (totally transparent to my ears with the standard profile). Vorbis and AAC fail together. Maybe a format limit. But the bitrate could be higher, and so the quality too... Need they more tuning ? For someone who like short_block test as music, certainly yes 


It's very difficult to concluded anything with pre-echo problem. Musepack is the winner, on natural and on artificial sample. Vorbis and PsyTEL aren't transparent in VBR mode under 400 kb/s. Vorbis fails on two point at -q10 on a critical electronic sample : pre-echo and VBR mode. AAC is not good at 170 kb/s (ultra) but is perfect on a false CBR 512 (250 kb/s).
Again :

1_ Musepack, clearly winner, more adaptative and « clever VBR »
2_ AAC, not transparent at ULTRA, but transparent on a strange CBR mode
3_ Vorbis, many difficulties with pre-echo (-q10 transparent with castanets, but not with electronic)

Code: [Select]
TEST No.1 : 



mpc 5 : /////·////

ogg 7 : **********

ogg 9 : **********

ogg10 : *****/****

short :      ·    



TEST No.2 : 



mpc 5 : //////////

ogg 7 : **********

ogg 9 : **********

ogg10 : **********

aac U : **********

aac512: /·//*/////

short :  ·


mpc5 : 1/20 [with ABC/HR, on many portions of the file]
ogg7 : 20/20
ogg9 : 20/20
ogg10 : 19/20
aacU : 10/10
aac512 : 1/10 - at one time, I gave original for compressed, and at 8 times, I wasn't able too hear a difference.
Wavpack Hybrid -c4hx6

MPC vs. ogg vorbis?

Reply #23 –
Quote
Originally posted by guruboolez
aac CBR 512 is only 250 kb/s... Ivan, what happened ?


It's impossible to understand AACenc's CBR.

MPC vs. ogg vorbis?

Reply #24 –
Quote
Originally posted by rjamorim


It's impossible to understand AACenc's CBR.


You are a specialist.
It's like wma, no ?
Wavpack Hybrid -c4hx6