Fraunhofer's MP3 Encoder Options

[Mats Peterson]

The latest MP3 encoder from Fraunhofer IIS does a very good job, and sounds much better than LAME in my book. It retains much of the depth and air in the music, while the LAME output sounds comparatively flat and lifeless. After all, Fraunhofer IIS has been a major participant in the development of the MP3 format.

[odious_m]

I must have a virus.

I clicked the link to hydrogenaudio.org, but was redirected here to joke_of_the_day.com.

[kornchild2002]

I must have a virus.

I clicked the link to hydrogenaudio.org, but was redirected here to joke_of_the_day.com.

I agree. It is perfectly alright to make such claims as long as one has multiple ABX test results to back it up. I am not talking about 10 or 12 ABX tests either. I am talking about hundreds of them. In order for such a claim to be made, one has to sample numerous tracks that represent one's music library.

Either way, this post wasted about 30 seconds of my time and is a joke. I am sorry but that is the way that I feel. These are all of the things that go into a blatant flame war thread:

1. Outrageous claim (check)
2. Lack of test results (check)
3. Vague language (check)
4. Lack of knowledge (check)
5. Lack of the acknowledgment of previous tests (check)
6. Overall smell of being a fanboy (check)

I can understand all this at it is your first post. So please go back, conduct some blind ABX tests (we can help you with that) with the FhG mp3 encoder and the Lame mp3 encoder, and then post back with your findings. Otherwise your post is about as useful as a poopy flavored lollipop. Please don't take any of this in a harsh manner though as I am simply trying to help you. Nearly all posts here on Hydrogenaudio are taken seriously, especially ones regarding the claims of audio quality of certain encoders.

[greynol]

Actually the proper test would be an ABC or ABC/HR.

http://en.wikipedia.org/wiki/Codec_listening_test

[Nick.C]

[O/T] In that Wikipedia link:

....
External links
Hydrogenaudio - Community audiophile site, host of most non-commercial ABX testing
....

Is that in the true sense of the word or the warped and scientifically blinded sense? [/O/T]

[halb27]

The latest MP3 encoder from Fraunhofer IIS does a very good job, and sounds much better than LAME in my book. ...

Fraunhofer mp3 encoders are expected to do a good job, as are Lame encoders. Which one to use is a matter of personal preference and may also depend on the way they are used and on what the focus is.
But there's also a strong phenomenon called placebo, which makes us easily think of a product being better than another one without any real justification. That's why blind abx listening tests are required to make sure that a certain assumption is valid.

What were your FhG and Lame settings with which you encoded your samples when comparing?
Are you willing to do a blind abx test to verify (or falsify) your assumption? It's easy to do with foobar.

[greynol]

Please, tell me how an ABX test can be used to determine which codec sounds better.

[Bodhi]

I must have a virus.

I clicked the link to hydrogenaudio.org, but was redirected here to joke_of_the_day.com.

[kornchild2002]

Actually the proper test would be an ABC or ABC/HR.

http://en.wikipedia.org/wiki/Codec_listening_test

Yes, you are right, that would be better. Still, a blind test is needed along with mountains of other information before their post can be taken with an ounce of seriousness. I am not trying to argue that the FhG encoder is bad. In fact, from my own tests, it can perform pretty good at high bitrates (but so can most other lossy encoders for that matter) of 192kbps and above. I am just trying to show that Mat needs to validate their claims by giving us more information and show us some blind ABC listening tests.

I can't vouch for your (loss of) hearing. All I can say is that the Fraunhofer encoder keeps much more of the depth and air in the music than LAME. Try it out yourself. I guess I don't need to add that you will need good equipment to hear the differences. Regarding lack of knowledge, I have a been a musician for 30 years, and I know what I'm talking about. This is no trolling. But if you want to keep that "whopeee, it's open source and we made it all by ourselves" attitude, fine.

Mat, please refrain from sending PM's like this. They are both insulting and take up space in my inbox. Terms like air are often used by airhead audiophiles and really describes nothing about the music. There is a big difference between being a musician and being one who has knowledge about lossy and lossless encoding. My post never said anything about Lame being open source so I don't know where you got that idea either. I wasn't trying to be negative in my post at all. Believe me, if I wanted to, I could have been insulting (such as the manner of your PM). I was simply stating that we need to see blind tests and we need to know more information about your claims. Otherwise I might as well claim that I am Jesus and anyone who doesn't believe in me will go to hell when the world ends in 2012. Do I have any proof for that? Absolutely not but it is fact and all who don't follow have (a lack of) knowledge. You see where I/we are coming from?

[Ron Jones]

air

Generally, whenever I see this word being used to describe sound, I start pulling my hair out. Whenever the word is used in the formation an opinionated statement that's phrased as if it were factual, I punch myself in the face.

Let's just say this thread hasn't been a fun one for me so far.

[kornchild2002]

The air comment was from a PM sent to me. I just thought that it would be fun to post the PM here instead of reply back to them through the private system. I know that the procedure of airing out the laundry in public is frowned upon here but this whole thread has been frowned upon.

Either way, I thought that the PM would show how Mat doesn't want to try to validate their statements and would rather (try to) insult me.

[simonh]

Mats is an idiot. Subject closed, surely?

How moronic that he looks at Open Source as a joke. Get a life boy.

This post has been edited by simonh: Aug 29 2008, 21:36

[Tahnru]

I've stated before, my assumption when I read posts like this is that they are from a viral marketing employee. Except if that were true, they're missing a big portion of the "formula".

http://www.penny-arcade.com/2006/01/27/guerilla-marketing/

[simonh]

Funny you should mention that Tahnru. I was picturing Mats in his Fraunhoffer cubicle firing that post off, at the behest of his superiors.

No doubt he is scurrying around other forums now, thankfully.

[greynol]

Most people I know who have been musicians for 30 years have suffered from moderate to severe high frequency loss. Not the best thing to tout when it comes to claiming acute listening skills if you ask me.

[halb27]

Please, tell me how an ABX test can be used to determine which codec sounds better.

Mats' description of FhG being 'better' was more or less 'identical to the original' when encoded with FhG but not with Lame. This can be proven or falsified by ABXing with respect to the tracks under investigation.

This post has been edited by halb27: Aug 29 2008, 23:10

[pdq]

I suspect that Mats has fallen into one of the "tweaking" traps, like forcing stereo mode, or disabling lowpass filtering. Perhaps that is why he has a low opinion of lame.

[AliL]

And so the bitching begins...

how about we close this thread and put it in the recycle bin before anything kicks off (too much) and when mats has conducted his ABC or ABC/HR tests he opens a new thread to reveal his results...

Just my 0.02

[Alex B]

I have personally tested two recent FhG versions against GoGo @ about 128 kbps (FhG: CBR & VBR, GoGo: ABR). I used 30 quite demanding samples. The results varied from a sample to another, but after all the encoders were statistically tied. GoGo is an old LAME version "on stereoids" (search for GoGo if you want to know more). I'd be surprised if the current LAME version would not be better.

I published the results in this post: http://www.hydrogenaudio.org/forums/index....st&p=574990

Since I was just preparing some images for another thread I can post a new presentation of the results:

I used the following codec settings:
FhGc = FIIS 3.4 CBR 128 kbps
FhGv = FIIS 4.1 VBR -br 0 -m 4 -vbri -ofl
FhGv1 = FIIS 4.1 VBR -br 0 -m 4 -q 1 -vbri -ofl
GoG = GoGo 3.13 ABR -a -b 137
GoG2 = GoGo 3.13 ABR -a -b 137 -q 2

Edit: I forgot to mention that I used FhG Mp3sdecoder v. 1.3 for decoding the samples. (Looks like the -ofl option worked. ABC-HR Java didn't adjust the starting position of the FhG VBR samples)

After seeing the results of my previous speed and bitrate test I upped the GoGo ABR bitrate setting from 135 to 137 so that it would better match the codec average. FhG VBR produced still a bit higher average bitrate with the now tested new samples (134 kbps vs. 131 kbps).

I tested 30 varied samples. All samples are created by myself and most of them are brand new. About half of them are from Vangelis' Alexander soundtrack album, which has a lot of interesting music (it's a mixture of classical, electronic and ethnic genres) and I wanted to try if that material would be useful for testing codecs.

The reference samples and my ABC-HR result files are available in this link:
http://rapidshare.com/files/55027379/alexb_test.zip (about 76 MB)

Here is a mirrow dl location that can be used if the rapidshare link does not work or is slow:
http://www.megaupload.com/?d=NFI54CWB

This post has been edited by Alex B: Aug 29 2008, 23:35

[Iceberg]

Being a registered user at hydrogenaudio for a whole year and publishing a first post that insults every devoted user in this site?

My LAME encodes are very deep and give me lots of air There, the last sentence evens the outrage against the LAME encoder.

[greynol]

Mats' description of FhG being 'better' was more or less 'identical to the original' when encoded with FhG but not with Lame. This can be proven or falsified by ABXing with respect to the tracks under investigation.

You'll see that he used the term "better" but certainly didn't use the term "identical". In fact, he said "retains much of the depth and air in the music," rather than saying it retains all of the depth as you are incorrectly implying.

[Alex B]

I would like to add that the purpose of my post was not to claim anything about the quality differences between FhG and GoGo. I did the test for a different purpose. I just happened to have the test results available and I wanted show an example of what kind of proof would make me to take a quality claim seriously.

When the test methods, samples and results are available others can post sample specific comments and the following discussion can actually lead to some new findings or mutually accepted conclusions.

Edit: added a missing "me"

This post has been edited by Alex B: Aug 30 2008, 00:38

[kiit]

Please, tell me how an ABX test can be used to determine which codec sounds better.

ah, but it could determine if he can even tell a difference reliably, no?

[Rio]

Not trolling here, but it is still not proven that LAME beats FhG at forced CBR 128.

Do I have to post ABX results when I post this? No.

[greynol]

You don't have to provide test results if you're simply asking a question.

ABX is useful in showing that there is a perceivable difference. It is not useful in showing that one is better than another unless one cannot be distinguished from the original, or it is much more difficult to distinguish one from the original than the other.

Again, the proper test in this instance is ABC or ABC/HR.