Hi peeps!

Well, it's been a glorious day here in the UK, so I've just sat down to have a pleasant evening sipping whisky and plundering my MP3 Jukebox on the home-built PC media centre.

I've been using Winamp as my jukebox front end for MPEG4 movies and MP3 (mostly high bitrate VBR) music. Bliss!

I can't actually hear anything wrong with the audio at all, but am I getting the best playback quality possible from my MP3 files?

So, am I missing anything? Are there any technically better decoders available?

Cheers, Slipstreem.

I used to use one called XMPlay, which claims to emphasize accurate playback of MP3s. It's the best-sounding I ever used, but I stopped using it because of the interface and started using WinAMP, even though it doesn't sound as good to me.

mpg123 is renouned to be a good decoder. This is the basis of the mpeg decoder used in foobar2000 and there is also a plugin available for Winamp (see below).

The current Nullsoft mpeg decoder is much better than earlier versions so you won't hear any difference IMO. But if you want to try out this decoder then get the following for Winamp:-

1) MPEG Decoder
2) Tag Reader (for the above decoder)
3) ASIO Output
4) 7-zip Archive (for opening the above files)

P.S. The ASIO Output has to be selected in preference to Directsound. It may not provide any audible difference in quality but it will mean that sound output from Winamp will not be processed by Windows first. Another alternative to try would be Kernel Streaming (<-- my fave) if you can't use ASIO.

Edit: Here is the Winamp Kernel Streaming plugin. I use foobar so I haven't actually tested this...

This is the same mp3 decode as exists in the present version of Winamp:
http://mp3decoders.mp3-tech.org/decoders_winamp.html

Thanks for the replies guys.

I followed up your links and did a little more surfing.

It would appear that Winamp's built in decoder is about as good as they get at present.

Maybe that's why I can't hear anything wrong with it. D'oh!

Thanks again, Slipstreem.

Too bad it doesn't support replaygain..

More here: Best sounding mp3 desktop player?

If you disable the 'Fast Layer' EQ options by unchecking the box in the MPEG audio decoder settings, it will fix the less than great mp3 sound issues in the default WinAmp settings. Now it should sound as good as any other great mp3 player out there.

Or don't use the equalizer.

That too but this is a separate EQ option in the mp3 decoder that is always on by default. IIRC WMP has a similar issue with mp3 playback.

It's actually rather buggy with MPEG 2 and MPEG 2.5 bitstreams.

I wouldn't recommend using it at all unless you are able to fix all these bug (like was done in foobar2000).

Since version 5.23 the Fast Layer EQ has been turned off by default.


I think it it the common view here, that most current decoders (foobar, winamp,...) offer excellent playback quality.

I also found another nasty bug with files encoded with the latest LAME:
http://www.hydrogenaudio.org/forums/index....8530&st=200

I can't believe in 2006 there's an audible difference between software decoders. An mp3 file decoded with different mp3 decoders should produce bit-indentical WAVs. Just done some tests and the results are far from bit-identical, and i'm extremely disappointed.

No. Rounding and dithering and what not can make data differ. No need to be disappointed.

If you actually measure the differences, they should be small.

That's good to hear. I've decided to stay pat with v5.11 because 5.2x doesn't like direct-x being disabled on my computers.

Foobar and Winamp both sound good to my ears.

That's the main reason that I used the Shibatch mpeg decoder in Winamp. Although, I'm sure Nullsoft will be adding replaygain support soon - right benski?

No they shouldn't. Google "floating point math".

Files don't have to be bit-identical to not have any audible differences (that should be quite obvious).

No harm intended to foobar2000 but - why weren't these mpglib modifications published, so the original mpg123 source would benefit of the bugfixes ? Maybe someone could politely ask ZzzZ (because afaik, it is not a requirement of LGPL) ?

It would be nice to have a source patch to finally squash these bugs...

Actually, it is a requirement of the LGPL to publish any modifications of software protected by the license. The purpose of the LGPL is so an author can write his/her program to utilize a library without having to publish the entire work under the GPL/LGPL, while all modifications to said library must be published under the LGPL.

However, according to the Foobar2000 license no open source components are used for mpeg decoding in Foobar2000.

Eh? The SDK contains the source to foo_input_std.

edit: It used to. "grep -ir mpglib ." and "grep -ir intput_std ." gives zero results. Was mpglib replaced?

Hmmm.... this is intriguing! Can anybody confirm if mpglib was replaced or if any other decoder is being used?

I wonder why there's no more upgrade of foo_input_mad for the new 0.9.x series of foobar2000?