so I ran across this

http://www.mars.org/home/rob/proj/mpeg/mad-plugin/

and am curious. It claims that it works by replacing the default mp3 decoder with an improved one. Has anyone tried it out? is it any good?

You'll hardly notice any difference in quality compared to the default plugin.

And in features, this plugin is (was) much worse. Last time I checked, it wouldn't read and write ID3v2 tags, and would steal all streaming being read by Winamp - no matter if it's mp3, vorbis, AAC...

I did a comparison at the beginning of this year. It's in french, but the final table should be understandable:
http://www.foobar2000.net/mp3decoder/conclusion.htm

Note that I've tested it on specific samples (very low volume samples, and very high high playback volume).
I can't ABX the quality progress on normal conditions.

there's no way those are correct, because they give foobar such a high score.. when foobar uses the same decoder as winamp

You are wrong. Very wrong.

And what about the MAD decoder inside Foobar?
What's the difference with the mpglib?

The decoder inside Foobar (the default one) is mpglib

Please, pretty please people, get informed before posting in this thread


Edit: OK, let me make you a list. For great justice.

Winamp: the default decoder is FhG. It can optionally use mpg123/mpglib and MAD.

Foobar: the default decoder is a modified version of mpg123/mpglib. IIRC, some decoding bugs were fixed and high quality dithering was added to Foobar's version. So, its quality is different to in_mpg123 for Winamp

Open your foobar and look at Standard Inputs (Preference, Playback, Input, Standard Inputs). You can choose between mpglib (default) and MAD.

Why oh why do these things only happen to me

Woohoo, here comes the clue train, last stop is you.

First, I said the default decoder inside foobar is mpglib. It is the default.

Second, I just installed the latest special edition installer, and here's what I got:
http://pessoal.onda.com.br/rjamorim/fb2kmp3.png

So you must be using an optional component God knows where from.

Third, I just know that foobar isn't using it by default - it isn't even installing it by default - because it is GPLd. That would force Peter to either release the entire player under GPL, or to license it from the developer for a healthy amount of cash, which makes no sense since mpglib is free and doesn't impose these limitations.

Oh well...

There is the possibility to use MAD inside Foobar and that was the purpose of my question.
I have found a small answer:
http://www.hydrogenaudio.org/forums/index....hl=mad+decoder#
I haven't seen any abx.

http://mp3decoders.mp3-tech.org/overall.html

very old decoder test

but it does support the fact that foobar2000 i better as it use mpg123
in the table mpg123 has more "blue=perfect" than mad


EDIT:
This is even ..........this table in this page will explain all your curiousity and doubt

Just to clarify, the standard MP3 input bundled with Foobar2000 uses mpglib at 32-bit floating point precision. Foo_mad uses libmad's standard precision of 4.28 fixed point, which is converted to the player's internal format of 64-bit floating point. All downsampling and dithering is performed by the player.

The Winamp MAD plug-in, as well as the console madplay utility, both use purely random noise shaped dithering, which merely rounds up or down based on a pseudo-random number generator, with no regard for rounding error. This, in my opinion, is worse than simple error diffusion. Or something.

Any theorical difference? No reason that one could be better?

Ooops double post
I'm in the train, so the connection is not fabulous

A bit off topic... Was Nitrane an own production, or was it based on something?

Edit: I am asking because that was the most crappy decoder I've ever seen.

Nitrane was comething code in house by Nullsoft, but PlayMedia sued because they though it contained AMP code. So eventually it was replaced with the FhG decoder.

Quality wise it amy have been bad, but it was one of the better ones for CPU useage, and when I was stuck on a 486 for too long that was nice.

According to some reverse engineering results done at the time by professional reverse engineers that testified on court, it actually had AMP code.

Well, all the better. AMP has always been a terrible decoder, so it was actually a good thing that Nullsoft was forced to switch to FhG

I'll repeat a point kode54 already made, but maybe isn't clear to marcan yet:
As opposed to Winamp, foobar2000 handles dithering in the core.
All input PlugIns decode into 32bit floating point, which is then replaygained, send through the DSPs and at the final stage converted to the output format. This is where the dithering happens. fb2k uses a highquality dithering implementation, which is far superior to the one of MAD (see guruboolez test for proof).
No matter if you use MAD or mpglib in foobar2000, you'll use its hq dithering and not MAD's crappy one. I doubt anyone will be able to hear MAD's slightly higher precision (ABX welcome).
The MAD Winamp PlugIn does its own dithering before sending the signal to the core.

I guess the only "real" advantage of MAD is its freeformat support which is missing in mpglib, IIRC.

No, foo_input_std's doesn't support free-format no matter what decoder is used.

Thx dev0.
I knew that the dithering is handled by Foobar but the precision of the MAD is higher. Difficult to hear though.

What makes you think MAD has higher precision?

offtopic, but lol :D

If you wanted to, you could make LAME encode MP3's at 640kbps and MAD would be able to decode them:

http://lame.sourceforge.net/USAGE

N.B. Search for "MAD", it's under "free format bitstreams"

That's pretty insane though, and I haven't tried it myself.

From the results of that ABX posted by guruboolez, I might switch to foobar.

I have steered clear of it until now, because it's a little bland in appearance and more confusing to operate the functions and equaliser (no presets like Winamp)

http://sjeng.org/ftp/fb2k/eq_presets.zip

Thanks, that kind of stuff will hopefully be included in future versions. There's a lot of extra crap thats downloadable for foobar, but thats what made me steer clear of it till now. You have to add extras to get the most from it.

Winamp's customisable too, but comes with a lot of prepackaged junk which, while generally unnecessary (like 20 included themes), is liked by people.

Is there a good site to go to for a 1 stop foobar plug-in/mod shop? I heard there was something that could add columns and all sorts of extras that I should probably experiment with

The special installer comes with the best/most commonly used 3rd party components.

From dev0:

Yeah but above that:



I'm not sure I'd say 32 bit fixed point is overall higher precision then 32 bit floating point, though I suppose it could be if values clustered very close to some value that was in the middle of the fixed points range. Certainly not any substantial difference between the two, not when its all decoded to 16 bit fixed point anyway.

Actual precision differs from 32bit floating point, which is just the format used internally by foobar2000.
menno tested it at some point and proved that MAD has a slightly higher precision.
Maybe the test could be redone at some point, but overall the real-world-difference should be nonexistant.

I think Windows Media Player 9 sounds better than both Winamp 5.05 and Foobar Special 0.83

Hello and welcome to Hydrogenaudio Forums,
Your post violates Terms of Service point 8:Please verify your claims, or your further posts on this topic will be removed by moderation, and appropiate actions will be taken against you if the problem continues.
Thank you.

It's very unlikely that Windows Media Player 9 sounds better than Winamp 5.05, since both use the exact same decoding engine (Fhg). foobar2000 may sound different due to dithering, but unless you provide ABX results this is irrelevant.

It's a joke!?

Not necessarily. WMP can include several DSPs (SRS, etc.). Usage of them can make the audio sound subjectively better, although, in that case, it depends exclusively on each person's tastes.

The MP3 decoder in WMP9 is also of quite good quality. It uses the FhG decoder, same as Winamp.

dev0: I gave an opinion, I did not make a statement. It would have been a statement if I'd said "Windows Media Player 9 definately better than both Winamp 5.05 and Foobar Special 0.83". Thanks for pointing that out though

I didn't know Winamp used the same decoder, but I use the equaliser in WMP and can generally make audio sound pretty good. It seems to come out louder (and sometimes) clearer/more vibrant than Winamp on my system for some reason.

I'm sure its most likely setting related. SRS and TruBass can sound good with certain tracks though. I'll consider following through some testing to work it out (though this seems silly if both have the same decoder, hence is almost guaranteed to be setting/DSP related)

it comes down to the same point in the end, it's just a matter of wrapping your statement in phrases. you might want to read the TOS.

that's totally subjective. some ppl might think that your eq settings sound unnatural, like crap, or like the best thing they ever heard.if you intersted, you can test for yourself what it sounds like if you have eq etc disabled in both wa and wmp. still hear a difference?

I think I located the source of the apparent poor quality of my files in Winamp. It seems the default setting of "Nullsoft's MPEG Audio Decoder 3.04" is "Fast Layer 3 EQ"

If I turn this off, the songs sound much better (cleaner and clearer)

An equalizer does neither have to do something with encoder nor with decoder quality. It's a tool to adjust playback for your equipment/preferences.
Making a statement about quality with an equalizer turned on is absolutely pointless.

I couldn't disagree more. Winamp's "Fast EQ" setting in the MPEG decoder was distorting the crap out of any lossy compressed file I played, even when it was set to "Normal" (i.e. everything flat).

I understand what you're saying about using it to determine overall quality of files, and I'm not disagreeing.

However the fact remains that when that particular setting was on, they sounded like shite, and this was happening to most of my files (hence it was of concern to me)

Well. It is just a good idea to turn the Fast EQ off. I wish that they had done this as a default setting. The fast EQ is just bad Equalizer-wise. The generic one is pretty good (written by one of the xmms authors), but only within new Winamp versions. The old generic Frankel EQ was pretty crappy.

But beware: Equalizing your music can make it sound more pleasing tto YOURSELF but technically it will rather harm the music quality (maybe cut frequencies as the old Frankö EQ did).

When you don't like the Winamp FHG decoder, have a look at otachan's in_mpg123 decoder since it is good featurewise AND qualitywise (32bit output, true gapless playing, dithering, hard limiting,...).

The only thing that it lacks is a decent id3 tag editor. If it had that, then it is perfect

Well, it links to mp3infp if you have that installed, and with the new Ver.2.47e it now also handles ape2 on mp3 files, which makes it start to become quite useable, actually...

Hi there!
I'll go a little off-topic but I think that this discussion is the right place...

I'm quite confused with MP3 decoders. From what I've read in the past I decided that the best way (for me) to decode my MP3's (LAME 3.90.3 -aps) is to use decoding through my front-end Monkey's Audio which uses the lame.exe for decoding.

Than I've read that according to MPEG Audio Decoder Compliance - Test Results MAD is the best decoder for all MPEG files so for my MP3's too. Then I started to read a MP3 Decoder Tests, that someone suggested and there's written "Every decoder should produce the same result" and also there's not MAD a winner...

My confusion is now a much bigger because of the above statement. A few years back I was decoding my MP3's and I was surprised that EAC's decoder produces different wavs than a certain version LAME, and this LAME produces different wavs than other certain version of LAME (comparing wavs in EAC)...

So where's the truth? Am I missing something...?

PS: What decoder is used by EAC?

That test was conduced by MAD's author. Take it with a grain of salt.

Please excuse reviving this dusty thread, but the relation is close:

Will there be a current version of "foo_mad" for foobar 0.9? I know people who love foobar2000, but can't play "freeformat" MP3s with the default decoder.

Kode54 made the MAD decoder long ago. But it is no longer available on his new page. I found the foobar2000 v0.8 compatible version from 2004 once by a deeplink somewhere else...