Okay, perhaps winamp is lying and the vbr of ogg is grabbing more variation of bit rates than it seems.

But then the file size would vary from file to file as much as mp3 or mp4. It doesn't.

First, I noticed 176 nominal to be far superior to r3mix settings on a quiete song (it took 171 abr to r3mix 151 abr), yet the loudest song 1 was weak sounding compared to the mp3 r3 mix settings (ogg=174 abr to r3mix=201). It took a 224 nominal (211 abr) to give the song enough punch to edge out the r3mix file.

Similarly, the mp4 directory files varied wildly in size, like the r3mix files.

Now, I thought the soft song surpassed the r3mix and the mp4 at an abr of 171, while the loud song did best at 211 abr. Obviously (because of practicality) I don't desire two profile drops for loud and soft files. There needs to be a vbr wide setting.

but you do get the concept of "nominal bitrate", do you?

sorry, couldn't resist.

Missing the boat here, aren't you?

The point is the vbr method is inferior to mp4 or simple r3mix settings. Soft songs sound too good, while a loud song sounds like crap. Unless you grab too many bits for every song.

Encoding some albums with -q5 (nominal bitrate of 160kbps), I get average bitrates ranging from 139kbps to 180kbps.

I don't know about VBR mp3 or mp4, but 40kbps difference seems pretty big to me..

To me, it looks like it's degarb who's missing the boat.

Codecs are for *listening*, not statistics.

Somehow I suspect that he/she's not listening very objectively.

Firstly, the quiete song DOES sound too good, while the Loud song sounds weak. This is a fact. Notwithstanding there is no objective listening, with anyone.

Secondly, file sizes do not lie. The mp4 dir is 54 megs, with wide variations on file abr. The mp3/r3mix dir is 59 meg with wide variation abr. The ogg/176 is 59 meg with little abr varation. The soft song files are too large, while the loud files are too small.

This is objective, yet, I only noticed this through listening test, not by looking at the abr or file sizes.

OGG will loose to mp4 hands down without a better wider vbr method. Hell, the mp3/r3mix loud song are more enjoyable than the ogg 176 settings. I point this out so they might fix it, rather than pretending this isn't a huge problem.

All testing done with optimus best head phones. (which belies the factual abr variations anyway.)

i'm still not understanding why you're using abr. is mp4 aac? i'm sorry, i'm just not familiar with what you're talking about...

AAC=mp4 afaik , since I used aac encoder and converted with winamp to mp4--likely the inventors (Dolby?) just bought the trademark, though fraunhoffer wrote some of the specs--I am guessing with the little aac reading I've done.

I was looking at the abr with alt3, not using abr to encode anything.

AAC is the audio stream. MP4 is the container (almost identical to Apple's .MOV container). FhG, Dolby, Sony, Nokia and AT&T were involved in the standardization process ("writing the specs").

And, AFAIK, AAC isn't a trademark.

The VBR modes in Vorbis are much better at handling certain situations (transients especially), but I've never heard a song sound "weak". The only reason why the bitrate doesn't increase as much on the louder songs is because vorbis handles these situations easily, whereas mp3 has difficulty and requires more bits. I'm not sure how the quiet song can sound "too good"...more detail on this from your point of view would be appreciated
thanks
B)

> I'm not sure how the quiet song can sound "too good"

Getting hung up on word "Too"
Meaning, more than adequate. In compression adequate is all we are aiming at.



Note that the Fraunhoffer vbr were bashed horribly by not taking a wide enough abr variation. Thus lame was touted the best vbr encoder out there--size v quality. This pattern too can be demonstrated in the fraunhoffer.

Is mp4 trademarked?

What do you mean? Patented?
If that's the case, you mean the container format or the whole technology portfolio?

AFAIK, the three charachters (MP4) isn't a trademark.

IF mp3/4 isn't trademarked, Ogg is stupid for not claiming the name mp5 Or perhaps, MPNext.

The very name ogg vorbis screams geekware, stay away normal people.

If you don't like ogg and are so in love with mp3 then just use mp3.

I personally have never heard any so-called "weak songs" produced by Ogg Vorbis and I listen to Death Metal, Black Metal, and Classical music. Even when the file size is some 500KB smaller than a LAME --alt-preset standard mp3 I've never experienced a "weak song."

Weak in relative terms to the r3mix lame, mp4, or the 224 nominal.

Weak meaning the song doesn't envelope you (as much in relative terms) like a warm blanket, as you get lost in the compexity of the sound. Not as much punch or interesting texture as the opposing files.

After all it takes texture to create music, not just a frequency range.

When I mention ogg vorbis to most mp3 users I know, they ask what kind of mouthwash or medicine is that.

LOL

http://www.hydrogenaudio.org/forums/index.....st=0#entry7182

Also, death metal, and black metal have few interesting textures to which to listen. And so, is not a good stick. I recomend test done with cleaned up lps with dj quality tables and needles. These, in theory, would be hard to encode, since they are still slightly dirty, and popfixed/noise reduced to point of death. Also, the music was plain more interesting in the days of lps.

First of all, where are you samples and abx tests degarb? Not only is it the policy of this board, but if you even expect people to take you remotely seriously you need provide this information.

Secondly, so-called "loud" songs cause problems with MP3 because MP3 cannot efficiently compress loud high frequency content (>16khz). I'm not going to explain why again, but if you want to understand why, you could easily find this information using the search function. Just look for sfb21, 16khz, etc.

Vorbis, AAC, and MPC do not suffer these problems. Admittedly I haven't experimented with AAC so much and so am not 100% familiar with it's nominal bitrate distribution patterns, but the fundamental problem which exists in MP3 does not exist in AAC.

Bitrates between different formats do not correlate to quality. You must use listening tests.

Perhaps a perfect example of this would be MPC, which also is similar in behavior to Vorbis in it's bitrate distribution. MPC --standard doesn't seem to "flex" much, but this doesn't mean anything at all in regards to quality -- the format is often considered the best there is moderate to high bitrates regardless of what it's bitrate distribution patterns look like. If anything, the fact that it does not flex in bitrate often only means that the psymodel is very stable and has few weaknesses in encoding different signals. This point can be further emphasized by Frank Klemm's work on SV8 and his plans to reduce bitrate jumps caused by impulse samples by using more efficient huffman compression, etc. When this is in place, MPC will seem to vary in bitrate even less, but still this will have no correlation with perceived quality.

FhG VBR wasn't said to be bad because it didn't flex enough (at least not by me), it simply just doesn't sound as good as LAME. This is verified via listening tests and has really nothing at all to do with bitrate comparisons.

Btw, if you aren't interested in performing blind listening tests, please don't continue this conversation and make claims like these. If you are interested though, please provide samples for people to download and blind test results. You can get testing software at http://www.pcabx.com or http://www.ff123.net

Have you ever in your life heard of Dimmu Borgir - nuff said!

Rhetorical question: What do you think your friends would have said if you told them you were making and/or listening to "mp3s" back in 1997?

As far as using "mp5" or hell, "mp" anything goes: It might not be illegal, but it'd be a spectacularly bad idea. If the MPEG Committee were to bless Xiph by proclaiming Ogg, Vorbis, Tarkin, etc. as MPEG standards, then that's a different story.

So your bottom line is ogg knows better at bit abr than aac? Personally, as stated I prefer the mp4 model better, with its wider range of abr. I feel the quiete songs could use less bits and the loud ones needed far more--when compared to the lame or mp4.

I noticed this through tests, then saw the abr.

I could easily provide tests, but there is way too much ogg bias here to cut with a knife.

>Perhaps a perfect example of this would be MPC, which also is similar in behavior to Vorbis in it's bitrate distribution. MPC --standard doesn't seem to "flex" much, but this doesn't mean anything at all in regards to quality -- the format is often considered the best there is moderate to high bitrates regardless of what it's bitrate distribution patterns look like.

Think about it, the lack of flex is the reason MPC is mostly good at high bit rates, rather than modest ones. Perhaps.


>FhG VBR wasn't said to be bad because it didn't flex enough (at least not by me), it simply just doesn't sound as good as LAME.

I wouldn't stake my life on that. Fraun did good 128kps, better perhaps, than lame in my tests. ( I could never get lame to do 128 as good my brother's MM 128cbr.) You just needed at least an 89% vbr fraunhoffer to get a r3 mix lame quality file. A larger vbr.

If someone denounces a tool in front its maker because it can't do something it wasn't designed for, or makes claims which they cannot really back up with hard objective data, or says people who strive to be unbiased are actually biased,

IMO I'd say that person is:

a. a clueless newbie who thinks he's not a newbie
or
b. a troll, posing as a clueless newbie who thinks he's not a newbie

Unfortunately, I think that degard falls under b.

He did not get a satisfactory answer in his first and only post here, and is now I believe trying to put .ogg down because he is bitter. Now maybe I am wrong on this but if you follow the two posts I don't beleve that I am. He first really wants to use .ogg but when he finds that he cant use replaygain to make his .oggs sound louder he then begins to bash it in this thread.

Just my thoughts.

No, I am in the camp, I want to believe in ogg, but let down on two counts. It is up to the ogg team to decide whether to address the problems or get deffensive. Getting defensive means the problem remains. If I didn't give a damn I would merely mosey off to mp4 and not bother typing or responding to idiots who defend rather than forward and investigate.

Who are you referring to with this comment?



Do you understand psychoacoustics? Have you ever read technical articles on how these encoders work? Have you ever actually worked on an encoder?

If not, then you're not really qualified to make statements about what could use "more bits" or "less bits", because you don't understand the technology the encoder is based upon in the first place, which means you don't really understand how the bits are spent or where they are going so to speak.



Who is being biased here? By not providing blind test results or test samples to prove your point, you're showing more bias than anyone else here. You refuse to back up your statements with reason or evidence which basically renders them as arbitrary, meaning that they must only be affected by personal preference and bias.

But degarb, you're not helping enough.

People here really do want hard data, really do want unbiased opinions, and certainly listen to constructive criticism.

I'd have to add option c and d.

c. a clueless newbie who doesn't want to troll, but does end up trolling
or
d. a newbie who wants to help, but doesn't know how... yet.

degarb, show us your powers. (As they say in Korea.)

The reason MPC is mostly good at high bitrates and not so good at low bitrates has to do with the technology the encoder uses -- that it is subband and not transform. It has nothing to do with the bitrate flexing. And about this bitrate flex, perhaps I should clarify. MPC actually flexes in its bitrate quite significantly, but it does so on a frame by frame basis. It is not uncommon to see the bitrate of MPC jump to 600kbps or sometimes even to 1000kbps or more for very short periods of time to encoder difficult signals. However, because the psymodel is very stable, the average bitrate does not vary much throughout a file or between different files.

Saying that MPC isn't good at low bitrates because it doesn't "flex" shows again that you just don't understand the technology.



Where are your test results and samples?

Also, making 1 correlation between one encoder and another encoder based on such unscientific testing principles ("I increased the slider and it sounded better so this must mean something in relation to all other encoders, etc") doesn't really prove anything meaningful overall.

LAME aps, perhaps the best mp3 vbr that exists, flexes less than most other VBR implementations, and yet it sounds the best. Again, this is because it (at least in effect) has a more stable psymodel. It does not dip too low, nor scale very high, except on impulse samples but then all worthwhile encoders scale up on impulse samples.

I'm sorry, but it appears that you are the one getting defensive.

So far you haven't provided a single shred of evidence to back up your claims. If they are as real as you claim they are, why don't you do this?

How do you expect people to improve upon a problem if they have nothing to work with? If anything, by refusing to provide this evidence, you just make it seem like you are either here to rant or troll and really have no interest in helping people to improve things, if there is even anything to improve at all!

I see this thread has deteriorated into a dogmatic religious fest. Time to end it.

I must have missed that part.

Looking up "dogma" in the dictionary we are greeted with this:

a point of view or tenet put forth as authoritative without adequate grounds

Hrmm. Well I guess I would have to agree with you that it sounds rather like "Ogg Can Be Inferior To Mp3 On Loud Music". Ok.

Unless... there are actual real grounds for this claim... but then, by copping out on us, it just doesn't seem that way, does it?

I can't really remember how many times I've said it at this point, but if people want to come here and post, and especially if they want to make claims which are both contrary to current knowledge and potentially inflammatory (i.e. they appear trollish), then they are expected to be reasonable, factually correct (at least as much as they can be), and able to validate their claims with at least some sort of evidence.

If on the other hand, you would rather simply argue over subjective observations and conjecture, there are plenty of other forums which would welcome that. You might even try usenet.

the earth is flat and i don't have to prove it.

Man!, if you want to be taken seriously, provide some clips, i could host them for you! But all you do is sort of trolling!

Several final points, and admonitions, before I get a real life and go back to my mp3s which play:

Unfortunately religious fervor for a product hurts that product more than helping it.

Actually, btw you can reasonable raise volume with replay gain =+3db and reducing the value for the track peak -.2. Yet knee jerk, a person who could never think of any reason he would want such a thing, refused to imagine a way to do this. Example of biased thinking.

You branded me a heretic the moment I said mp4 was the superior format. You started squealing , rather than do some tests your selves. Then desire me to host 10 megs and bit downloads for free to boot.

Also, just because a file sounds closer to the wav DOES NOT mean it sounds better. You can improve on the human acoustical model to make things sound more humanly desirable. So don't get that confused. Lame is the nasal of encoders compared to the spectum truer wma, but people advocated it religiously.

I bet the same people attacking me here are waiting for the second coming of Torvalds, who will finish Linux with all needed drivers and properly written help files in real English. (Ouch that hurt to type; it was so heretical. See resulting f-ups.)

Which has thoroughly convinced me that degarb's worse than a good for nothing troll. Sorry guys, I always fall for flamebait.

And which ... ummm... pharmacueticals ... helped you get 'lost'?

Lock thread please. We've wasted enough time on this idiot already.

What you need is a sound compressor. Search for one on Google. And louder doesn't mean better... I usually use mp3gain at the 83dB setting

You got flamed because:
1) the title of the thread
2) You're saying just lots of unfundamented bsh*t

Anyway, I must say that blade mp3 encoder, besides Tonal Purity™, also ads the Warm Blanket™ feeling to the original wave file... LOL

Hmm..
ogg can be inferior to mp3 with loud music - true
mp3 can be inferior to ogg with loud music - true
ogg can be inferior to mp3 with quiet music - true
mp3 can be inerior to ogg with quiet music - true

These all cases can be true depending on the case. Which happens more often, which is the trend, that's the question which I'm not trying to answer or bother even giving my opinion at this point.. . One thing what one can't do is compare vbr bitrates and proclaim mp3 sounds better than another codec because of higher vbr bitrate on some music.
When comparing different codecs and even different codec settings, simple "higher bitrate with codec/setting A -> always better quality than lower bitrate codec/setting B" is simply not valid. It's not so simple at all.
This is especially so when comparing true vbr codec to a codec like mp3 which has all kinds of different problems with vbr (some can be made better with proper tweaking).

Anyway, since this thread doesn't seem to go any further, and the original poster doesn't want to give one example (btw. I'm sure he could find several examples ), and this thread starts to be mainly ranting, I'm closing this thread now.