Great Mp3 decoder in terms of sound quality – Mpadec 1.0.0.0

[aquamarinebeach]

I find that the old versions of Mpadec (i.e the ones written by Gabest) are the best mp3 decoders even today in terms of sound quality on my "relatively high-end audio setup". The new versions released in MPC-HC have gone down in quality.

Some people have asked me to give some details on my sound setup:

A) The one I am testing is (my computer room setup) - i.e. my M-Audio Revolution 7.1 sound card converts stereo sound into 5.1 analog sound using SRS Circle Surround II that's built right into its drivers. The Soundcard drivers themselves take care of bass redirection / crossover to a Klipsch Promedia Ultra 5.1 Multimedia Speaker System.
This is why the volume is being controlled from my PC - as in this case my sound card is my receiver and the Klipsch speaker system has its in-built amplifier.

B) The other sound setup (my living room) uses the Digital Spdif port of the M-Audio Revolution 7.1 sound card. Sound is sent to a Yamaha RX-V1900 7.1 Channel Home Theater Receiver. The receiver uses SRS Circle Surround II (yes I love that technology) to output the sound to a Boston Acoustics A 2310 HTS 5.1 Home Theater Speaker Package.

And the reason why I started this thread and am so impressed with the internal mp3 decoder of MPC 6.4.8.6 is because it makes (Sound Setup A) sound very close to (Sound Setup B)
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Some people have mentioned the difference between the mp3 decoders should be negligible.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

However, regarding the difference in sound quality between MPADec 1.0.0.0 and the other ones - for me it's more than subtle - but then may be it's because I am listening to them on 5.1 systems. The sound stage is therefore bigger and therefore the differences are more discernible. I am sure if I were to listen to them on stereo speakers or headphones, then I too would not find much of a difference between the decoders.

I just discovered their awesome quality by accident a few weeks ago - and it is way better than the better known Mp3 decoders that I have been using until this great find (i.e. Apollo, Foobar2000, KMPlayer, Winamp…).

I have found mpasplitter.ax v 1.0.0.2 to be the best in terms of splitter.

P.S: It's in the wrong (MP4 filter folder on sourceforge) Download URL: https://sourceforge.net/projects/guliverkli/files/MP4%20Splitter/MPA%20Splitter%201.0.0.2/

As for the mpa decoder, I have found the one in MPC 6.4.8.6 - i.e the last build of mpa decoder v 1.0.0.0 to be the best sounding - huge soundstage, clear lyrics and tight bass.

The second best is the last Gabest version - i.e. mpadecfilter.ax v 1.0.0.4. While it is good, the sound is boomier, the bass is less tight and the lyrics are louder but more shrill compared to the perfect decoder of MPC 6.4.8.6 (latest version of MPC with that decoder - ver 6.4.8.7 switched to mpadecfilter.ax v 1.0.0.1 and the sound quality degraded)

I find that the even more recent builds of mpadecfilter.ax (now based on FFmpeg) are even worse in terms of audio fidelity - the soundstage is poor when compared to the old versions...)

Right now I am stuck with using MPC 6.4.8.6 (very obsolete and buggy player) and using mpasplitter.ax v 1.0.0.2 loaded externally, combined with the internal decoder - simply because the sound quality is out of this world.

I have not been able to find the stand-alone filter mpadecfilter.ax v 1.0.0.0 in order to use it in more modern and sleek directshow players like Imesh (excellent interface, features and organization of my whole mp3 library and uses mpadecfilter.ax - albeit a later MPC-HC build. This file can of course be overwritten by any other version of mpadecfilter.ax )

It’s very unfortunate that the sourceforge URL of Media Player Classic has the later builds but is missing the oldest but best one.

Namely, it has has MPA Decoder 1.0.0.1, 1.0.0.2, 1.0.0.3 but it is missing the stand alone filter MPA Decoder 1.0.0.0.

Does anyone happen to have the stand-alone filter mpadecfilter.ax v 1.0.0.0 by any chance? That would be a real ife saver indeed.

Or is someone computer savvy enough to build an mpadecfilter.ax v 1.0.0.0 from the source code of MPC 6.4.8.6?

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

*****UPDATE*****: for those who want to give a stand-alone version of this decoder a try:

All my respect and thanks go to to [REX] for compiling an excellent Mpadec 1.0.0.0 from the SVN 447 (Source Code of MPC 6.4.8.6)

http://guliverkli.svn.sourceforge.net/viewvc/guliverkli?view=revision&revision=447

He compiled it using the June 2010 DirectX9 libraries - it sounds just like the internal decoder of MPC 6.4.8.6 - only difference is it is a bit boomier (due to the June 2010 DirectX libraries I presume) - to my ears that makes it lose a bit of the accuracy in the excellent soundstage of MPC 6.4.8.6

If I rate MPC 6.4.8.6's internal Mp3 decoder (10/10), I would rate [REX]'s Mpadec 1.0.0.0 (9/10). They are very close in terms of note rendition.

Feel free to download it @:

http://rghost.net/43433617

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I am now impatiently waiting for some generous soul to compile Mpadecfilter.ax 1.0.0.0 from SVN 447 (Source Code of MPC 6.4.8.6)

http://guliverkli.svn.sourceforge.net/viewvc/guliverkli?view=revision&revision=447

and October 2005 Directx 9 Libraries (the decoder will sound the best with those as MPC 6.4.8.6 was built around the latest DirectX libraries available at the time of its release - i.e 6 Nov 2005) - Download URL below:

http://filemare.com/browse/115.90.232.68/driver/dxsdk_oct2005.exe

P.S: MPC 6.4.8.6 was compiled with VB2005. My knowledge about compiling is very limited. So, I don't know whether using VB 2005 or VB 2008 could also make a difference to the way the decoder sounds.

P.S: JUST A HEADS-UP: I have read that it's best to uninstall the current version of the DirectX SDK before installing another version in order to avoid incompatibilities / file conflicts. Of course once you are done using the older DirectX SDK, you can uninstall it and revert to the latest DirectX SDK.....

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Come on people, if you really like the way this internal decoder sounds and happen to know how to compile source code, then please consider compiling the above decoder for our enjoyment.

Thank you in advance,

Ray

[wanezhiling]

Did you try LAV audio decoder?

[tetsuo55]

Memory is foggy, but i think we switched over because there was a pretty bad bug in that old version of mpa that distorted the sound quality, it was clipping stuff or doing something worse.

Maybe you prefer the sound created by this distortion? have you tried using noise sharpener in foobar2000? it creates a similar effect.

[aquamarinebeach]

Memory is foggy, but i think we switched over because there was a pretty bad bug in that old version of mpa that distorted the sound quality, it was clipping stuff or doing something worse.

Maybe you prefer the sound created by this distortion? have you tried using noise sharpener in foobar2000? it creates a similar effect.

tetsuo55 - could you please check to see if you still have a copy of the mpadecfilter.ax v 1.0.0.0 file, since you worked with it? That would be a life-saver

Lol interesting - I do not think I would enjoy clipping or distorted sound - on the contrary this version plays my mp3's less loudly (compared to later builds) which is party why even at deafening volumes it remains clear sounding.

What's interesting is that MPC 6.4.8.6 sounds different (i.e better than MPC 6.4.8.3, MPC 6.4.8.4 and MPC 6.4.8.5) - which all have mpa decoder v 1.0.0.0. So I am guessing v 6.8.4.6 uses a later build of that decoder - i.e the last one. And then the sound quality did not stay as good in the following release - MPC 6.4.8.7 (which uses mpa decoder v 1.0.0.1) That's the first version of the stand-alone mpa decoder that's available on sourceforge

Let me give the foobar plugin a try and I'll let you know

wanezhiling - nope - I have used it for movies before - I'll test it out for Mp3's and give you my feedback.

Thanks for your replies guys

[filler56789]

I find that the old versions of Mpadec (i.e the ones written by Gabest) are the best mp3 decoders even today in terms of sound quality on my high-end audio setup.

Seriously? :confused:

Have you ever tried the mpg123 DirectShow filter? You can configure it to ignore :cool: the ReplayGain info, for example. :)

[tetsuo55]

On closer inspection it looks like these builds are from before my time.

I did check out the changelog which you can view here http://guliverkli.svn.sourceforge.net/viewvc/guliverkli/trunk/?view=log

I did not find anything that would point clearly to a change in audio quality.

Some googling brought me to this:
http://www.hydrogenaudio.org/forums/index.php?showtopic=98379
and this:
http://www.hydrogenaudio.org/forums/index.php?showtopic=98579

Those links are discussions about differences in audio quality including talk about how to do a "proper" comparison test. Even when you hear differences they might be due to bugs causing the music to sound better subjectively.

[aquamarinebeach]

Seriously? :confused:

Have you ever tried the mpg123 DirectShow filter? You can configure it to ignore :cool: the ReplayGain info, for example. :)

filler56789 - I have tried mpg123 way back in 2005 or something - are you referring to the ones available for download @

http://mpg123.org/download/win32/

I gotta give those a try

[filler56789]

^ More or less. That Sourceforge page is about a standalone MP-1-2-3 :) player, whereas I'm talking about a DirectShow source filter and a DirectShow decoder.

mpg123 DirectShow Filter 1.045.exe @

http://softvisia.com/users/Night_Raven/CodecFAQ/Filters/Audio_filters/

1) To whom this may interest, the installer can be opened with 7-zip

2) To whom this may concern ;) , the installer does include the respective source-code :sly:

[leeperry]

"best" as in can you show us the improvement somehow? or is that a subjective impression? did you ensure that they were all outputting 32fp?

[aquamarinebeach]

"best" as in can you show us the improvement somehow? or is that a subjective impression? did you ensure that they were all outputting 32fp?

Although my sound card can do 24-bit processing I find that upon playing Mp3's - i.e. 16-bit files, using a higher setting like 24-bit or 32-bit adds artifacts to the sound due to the upscaling of the original resolution.

Now when I watch HD-Movies in PowerDvd that have a 24-bit DTS-HD soundtrack, that's a different story. I make sure the sound card is doing 24-bit decoding all the way.

[aquamarinebeach]

Armada – I got more time to test out the difference in sound between the internal decoder of MPC 6.4.8.6 and the external decoder you compiled for me.

You were right. They have the same source code. From my listening tests, they sound very similar in terms of rendition of the notes. The only difference is that the one you built sounds boomier and a bit less accurate than the internal decoder.

Since I made sure to use the exact same splitters and used both decoders in MPC 6.4.8.6 along with the same DirectX output renderer, the only remaining culprit for the difference would be the DirectX SDK versions used in compiling the 2 decoders unless of course if using VB2005 or VB2008 could cause a difference too?

As you already know I do not have a whole lot of knowledge on the above 2 applications – but I did read that different sdk’s can affect the compilation.

From the MPC release infos:

Changes in Media Player Classic 6.4.8.6 for Win2k/XP (the date on MPC folder files is 6 Nov 2005)

- d3dx9_27.dll is a separate download now because it is no longer needed to run mpc, however to use pixel shaders it has to be put in the dll path.

From:

http://www.toymaker.info/Games/html/d3dx_dlls.html

and

http://www.toymaker.info/Games/html/directx_9_0c.html

and since that MPC version was released on 7 Nov 2005 and d3dx9_27.dll was released as its accompanying download – it can safely be deduced that either:

DirectX 9.0c (Aug 05)

http://www.filehippo.com/download_directx/260/

or

DirectX 9.0c (Oct 05)

http://www.filehippo.com/download_directx/479/

was used to compile MPC 6.4.8.6

based on:

DirectX SDK October Update 2005

Changes in this update:
• XInput - a new API for the Xbox 360 controller
• XACT beta - cross platform audio creation tool
• Improved support for Visual Studio 2005
• Effect Edit and Mesh View tools removed. The new DirectX Ops and DirectX Viewer tools are meant to replace the functionality however they do not have all the previous functionality so you may wish to keep the tools from the old SDK.

DirectX SDK August Update 2005

Only a few small changes this update:
• PIX (Performance Analyser for DirectX) tool update
• Effect system updates
• New sprite sorting flags.
• The D3DX DLL is now named D3DX9_27.dll (see note)
• DirectX Viewer - allows viewing x files with DXSAS compliant effect. As this tool is replacing functionality currently in MView (for viewing mesh) and EffectEdit (editing and viewing shaders and effects) these two tools are now deprecated and in fact MView will be removed in the next SDK release..
• A few other minor changes (see the DirectX SDK Download Page for more details).
• The sample browser will not work with IE 7 !
Note: DirectShow was moved to the Platform SDK in the April release and overall DirectX is becoming more reliant on the Platform SDK, so always a good idea to get the latest.
Finally, from the following it is safe to deduce that MPC 6.4.8.6 was compiled with VS2005

Changes in Media Player Classic 6.4.9.1 :

- update svn ignore list;
- add the needed DirectX include files and ddraw.lib;
- add the atl include files;
- add VS2008 project/solution files (quickly updated from VS2005 files with the minimum needed changes to compile);
- update common/release.vsprops and the manifest file;
- now mpc should compile with VS2008;
- backport r3013,r3019 from mpc-hc svn.

In the light of the above, Armada could you please please please recompile 2 external decoders for me since you are so smart and it only took you a few minutes the last time.

1) Using DirectX SDK August 2005 + VS 2005

2) Using DirectX SDK October 2005 + VS 2005

Since the sound is already so close, I am sure either 1) or 2) above will replicate the sound of the internal decoder of MPC 6.4.8.6.

Hope my analysis is objective this time and not based on "magic" - lol!

And I promise to never bug you again about this decoder…LOL
Thanks in advance for your help,

Ray

[Armada]

I already told you that's impossible, those SDKs no longer exist. The links you've provided point to the DirectX runtime, not the SDK.

I will not spend anymore time on this based on conjecture. People have given you plenty of advice on how to prove your suspicions, but so far you have yet to produce a single proof that would indicate there even is a difference.

[filler56789]

Although my sound card can do 24-bit processing I find that upon playing Mp3's - i.e. 16-bit files, using a higher setting like 24-bit or 32-bit adds artifacts to the sound due to the upscaling of the original resolution.

If you find the time to read many posts from madshi about Digital Signal Processing, you'll realize that what you said doesn't make much sense. :)

[nevcairiel]

Especially because most MP3 decoders internally work in floating point anyway, and an output in 16-bit integer actually cuts away information (MP3 is not "16-bit files", they are lossy, so they don't have an inherent bitdepth)

[iSunrise]

I find that the old versions of Mpadec (i.e the ones written by Gabest) are the best mp3 decoders even today in terms of sound quality on my high-end audio setup. The new versions released in MPC-HC have gone down in quality.

I just discovered their awesome quality by accident a few weeks ago - and it is way better than the better known Mp3 decoders that I have been using until this great find (i.e. Apollo, Foobar2000, KMPlayer, Winamp…).
How did you test Foobar2000, with the included newer ffmpeg based decoder or an older version? Also, did you use it´s default output settings or installed the additional WASAPI component and chose the bitdepth according to what your soundcard supports.

What would also be quite helpful is if you could describe your audio setup in more detail, especially what soundcard you are using and how it´s connected to your receiver/amp.

[aquamarinebeach]

I already told you that's impossible, those SDKs no longer exist. The links you've provided point to the DirectX runtime, not the SDK.

I will not spend anymore time on this based on conjecture. People have given you plenty of advice on how to prove your suspicions, but so far you have yet to produce a single proof that would indicate there even is a difference.

Armada - I tried to use the Haali Muxer with the internal decoder of MPC 6.4.8.6 as per your suggestion but it did not work - it only worked with the external decoder you created for me.

Do you know of any other methods that I can use to show that there indeed is a difference (as they definitely do sound different)

Now if you could create another decoder using the October 2005 SDK, I will be more than happy to demonstrate that Haali does indeed give different results.

I was able to find the SDK at the URL below

http://filemare.com/browse/115.90.232.68/driver/dxsdk_oct2005.exe

Also, just to confirm - MPC 6.4.8.6 (released on 6 Nov 2005) uses the code from SVN 447, right?

http://guliverkli.svn.sourceforge.net/viewvc/guliverkli?view=revision&revision=447

Thanks for your help in advance man

[filler56789]

Do you know of any other methods that I can use to show that there indeed is a difference (as they definitely do sound different)

As a last resort, you still can record "What-U-Hear" with Goldwave, Audacity, Wavosaur, whatever.

I was able to find the SDK at the URL below

http://filemare.com/browse/115.90.232.68/driver/dxsdk_oct2005.exe

Congratulations and my respect for the HUGE perseverance. :)

[aquamarinebeach]

As a last resort, you still can record "What-U-Hear" with Goldwave, Audacity, Wavosaur, whatever.



Congratulations and my respect for the HUGE perseverance. :)

Thanks filler - that file was indeed very hard to find.

You seem to know your stuff about Audio Applications.

Just wondering if you got to test out MPC 6.4.8.6 for playing mp3s?

[leeperry]

MPC's mp3 decoder is using libmad IIRC, and it's also available in ffdshow.

You could always run it through this test: http://mp3decoders.mp3-tech.org/24bit.html

[filler56789]

Just wondering if you got to test out MPC 6.4.8.6 for playing mp3s?

Even when I could use MPC on my machines, I always disabled all internal filters, so no, I never knew how good or how bad MPC's internal MP3 decoder was. :o In fact, I did prefer MPC to MPC-HC, but only until I found out it behaves "buggily" :) under Windows XP Pro with Service Pack 3. :confused:

Anyway, for listening to MP3 files, I've always used and preferred Winamp (in the past) and foobar2000 (in the present). For the DirectShow players, I chose the mpg123 source and decoder because they are highly-configurable, and made me be sure that a FP 32-bit output is the way to go ( even though the decoder allegedly is capable of FP 64-bit processing :eek: :) ).

[aquamarinebeach]

filler56789 - I used to be the same way - I always disabled the internal decoders and was using decoders from PowerDvd 10 for e.g. I am using Win Xp Pro SP3 too - (looks gorgeous with the Zune Theme Color Mod by atomikscr) - love the Black Color compared to the very ugly Blue and Silver colors in the WinXP.

Yeah talking about buggy - MPC version 6.4.8.6 is indeed very buggy - for e.g. the Shuffle button does not work. But the reason I am sticking to it for now is because on my system - the sound is just out of this world for Mp3's.

I have used Winamp in the past too - till I started hating the bass outputted by the Fraunhofer decoder - very boomy and not tight at all on my setup.
On my system Foobar2000 sounds too flat for my liking.
I tried Apollo - similar sound to Foobar2000 - a bit better I find - but still too flat and very old interface.

The best software I have found for mp3's - i.e sleek and comprehensive and very light software that does not hog system resources is Imesh 11 (organizing by artist, song title, bitrate,....and for extras can even play youtube videos inside the player, as well as it's got a DJ feature where you can type in a song and it will play similar songs from the web for you....

Just some mods done - deleted ac3filter.ax and ammp3.dll (as it's an amplifier and adds coloration to the file). Also delete the mpadecfilter.ax file as it uses a later build from MPC-HC 1.2 and replaced it by one of the earlier Gabest versions). It's a directshow player so basically you can use mpg123 in it too if you wish - by changing merits in Radlight filter manager.

Give it a try. I am sure you will love it.

Regarding 64-bit processing apparently it does not exist as per the link above Leeperry provided: http://mp3decoders.mp3-tech.org/24bit.html

As for me I find that when I use 24-bit or anything higher for Mp3's I hear artifacts in the sound (extra bits that aren't supposed to be there in the first place???) - that make my ears ache - so I just stick to 16-bit decoding mp3's.

Pls give the internal decoder of MPC 6.4.8.6 a try for your Mp3's. They work best with MPASplitter 1.0.0.2 - which happens to be in the wrong folder i.e MP4 splitter folder on sourceforge https://sourceforge.net/projects/guliverkli/files/MP4%20Splitter/MPA%20Splitter%201.0.0.2/

I would really love to know your SUBJECTIVE impressions about the sound on your sound setup

Regards,

Ray

[leeperry]

it's nice and all but MPC won't do gapless playback, and MP3's are so yesterday anyway....lossless all the way :)

[aquamarinebeach]

I fully agree - however 90% of free songs are still mp3's

[iSunrise]

I fully agree - however 90% of free songs are still mp3's
Yes and it sucks. Even iTunes doesn´t offer any lossless options. I would have bought a shitload of albums already (I have like 80 albums on my wish list) if they were at least ALAC or FLAC. Since that is still not available anywhere with a decent library I still have to buy CDs.

[leeperry]

what? ;)

anyway, the OP might as well use libmad in ffdshow and output 32fp = problem solved. But IIRC some VBR files wouldn't decode properly and I had to use libavcodec.

[aquamarinebeach]

what? ;)

anyway, the OP might as well use libmad in ffdshow and output 32fp = problem solved. But IIRC some VBR files wouldn't decode properly and I had to use libavcodec.

Leeperry - what source/splitter are you using to decode the mp3's?

I am currently testing out the ffdshow audio decoders. Does ffdshow not include an mp3 source/splitter? Cannot locate one using Radlight filter manager...

So far the major difference I am noticing is that compared to the well-balanced sound the internal decoder of MPC 6.4.8.6 produces on my system, the ffdshow decoder's sound is overly bassy (both with 16-bit & 32-bit floating point)

It would be great if you could compare the sounds on your audio setup as well...

I am using MpaSplitter 1.0.0.2 and so far all the mp3's I have tested decoded fine with IIRC.

P.S: It's in the wrong folder (MP4 Splitters on sourceforge)

https://sourceforge.net/projects/guliverkli/files/MP4%20Splitter/MPA%20Splitter%201.0.0.2/

[filler56789]

Does ffdshow not include an mp3 source/splitter? Cannot locate one using Radlight filter manager...

ffdshow is a set of decoders and processing filters. It has never included source filters or demultiplexers (media splitters).

So far the major difference I am noticing is that compared to the well-balanced sound the internal decoder of MPC 6.4.8.6 produces on my system, the ffdshow decoder's sound is overly bassy (both with 16-bit & 32-bit floating point)

ffdshow can use either libmad or libavcodec. Do you mean they both suck in the same way and degree? :) :D

Regarding 64-bit processing apparently it does not exist as per the link above Leeperry provided: http://mp3decoders.mp3-tech.org/24bit.html

That page looks somewhat outdated to me. ;) FWIW, the config applet of the mpg123 filter lets you choose a 64-bit FP output, but all options except 32-bit FP make the filter quite unstable on my XP machine. :confused:

As for me I find that when I use 24-bit or anything higher for Mp3's I hear artifacts in the sound (extra bits that aren't supposed to be there in the first place???) - that make my ears ache - so I just stick to 16-bit decoding mp3's.

More probably, the higher bitdepth shows you how much most MP3 files suck :sly: , and you're just trying to deny the harsh reality. :D

Pls give the internal decoder of MPC 6.4.8.6 a try for your Mp3's.

I prefer to wait until someone builds for you a standalone filter according to your recommendations (http://forum.doom9.org/showthread.php?p=1614051#post1614051) --- so I could test in the Graphstudio "environment". As I said before, mplayerc.exe is rather unstable under XP Pro w/ SP3, and it MAY cause very-weird interferences on other applications and on the OS itself. :(

[aquamarinebeach]

filler56789 - so the processing filter of ffdshow would control the equalizer, spatial effects,...i.e post processing? I tried to see if that filter would work as a splitter and it did not...

Impression on ffdshow audio decoders on my system (subjective):

libavcodec - I just hate the bass (not 3dimensional enough for me) and lyrics are too harsh

libmad: 3dimensional bass and lyrics more subdued.

I prefer libmad to libav. However, MPASplitter + internal decoder of MPC 6.4.8.6 produces less boomy bass than ffdshow libmad (i.e more tight and accurate sounding). Also lyrics and bass are enveloped together - i.e. bass and lyrics do seem more separate in ffdshow.

Same thing here with the mad+mpg123 plugin u referred me to, only 32bit floating works on my WinXP machine and just like for ffdshow, I prefer the sound from mad over here too.

By the way on my system, MpaSplitter 1.0.0.2 works better with the mad+mpg123 decoder than the included splitter (nicer sound and more compatible with mp3's)

"More probably, the higher bitdepth shows you how much most MP3 files suck , and you're just trying to deny the harsh reality"

You are probably right - but I am all about making mp3's sound the best they can on my system given their limitation - I am not looking for the most mathematically accurate sound reproduction - just the most pleasant to my ears.

After all they are likely to remain the defacto standard for a long time - given the much huger size of lossless files.

"I prefer to wait until someone builds for you a standalone filter according to your recommendations --- so I could test in the Graphstudio "environment". As I said before, mplayerc.exe is rather unstable under XP Pro w/ SP3, and it MAY cause very-weird interferences on other applications and on the OS itself."

It's your lucky day - big credits go to [REX].

He compiled an excellent Mpadec 1.0.0.0 from the SVN 447 (Source Code of MPC 6.4.8.6)

http://guliverkli.svn.sourceforge.net/viewvc/guliverkli?view=revision&revision=447


He compiled it using the June 2010 DirectX9 libraries - it sounds just like the internal decoder of MPC 6.4.8.6 - only difference is it is a bit boomier (due to the June 2010 DirectX libraries I presume) - to my ears that makes it lose a bit of the accuracy in the excellent soundstage of MPC 6.4.8.6

If I rate MPC 6.4.8.6's internal Mp3 decoder (10/10), I would rate [REX]'s Mpadec 1.0.0.0 (9/10).

Feel free to download it @:

http://rghost.net/43433617

And I am also curious to see what your GraphStudio tests reveal...please share your findings.

P.S: I find it works beautifully with MPASplitter 1.0.0.2

https://sourceforge.net/projects/guliverkli/files/MP4%20Splitter/MPA%20Splitter%201.0.0.2/

You are right about mplayerc.exe...

Strangely, when I tested Mpasplitter 1.0.0.2 & Rex's Mpadecfilter 1.0.0.0 in different versions of MPC - MPC 6.4.8.6 gave me the best sound.

For e.g at the same volume settings, the combo sounded louder and less clear in MPC 6.4.9.1 (build 114). May be it's got something to do with the fact MPC 6.4.8.6 - has it's own page where you can adjust the decoder settings (i.e. 16 bit, 32 bit, floating....) i.e. MPAdecfilter 1.0.0.0 cannot be right clicked to change the bitrate....

While in MPC 6.4.9.1, the decoder settings were moved on to the decoder itself, i.e. MPADecfiter 1.0.0.2 onwards can be right clicked to change the bitrate....

When trying the above splitter/decoder in MPC-HC versions, the volume was boosted even more giving out a harsher sound. (Changes were done to the mixer between MPC and MPC-HC versions??? I wonder...)

---------------------------------------------------------------------------------------------------------------
Impatiently waiting for some generous soul to compile Mpadec 1.0.0.0 from SVN 447 and October 2005 Directx 9 Libraries (will sound the best with those as the decoder was built around the latest DirectX libraries available at the time of the release of MPC 6.4.8.6 - i.e 6 Nov 2005) - Download URL below:

http://filemare.com/browse/115.90.232.68/driver/dxsdk_oct2005.exe

P.S: MPC 6.4.8.6 was compiled with VB2005. My knowledge about compiling is very limited. So, I don't know whether using VB 2005 or VB 2008 could also make a difference to the way the decoder sounds.

----------------------------------------------------------------------------------------------------------------------

filler56789 - do you by any chance have the skills to compile this decoder?

[bjd]

Always felt the move from 16bit -> 32fp processing for the ffmpeg decoders improved the sound quality in LAV.

I don't know what your exact set up is (aquamarinebeach), but avoiding direct sound at all costs and ensuring a bit exact flow to your high end system will give you the best mp3 playback regardless of decoder - Do you us ReClock in WASAPI mode ? I would recommend it to anyone serious about audio playback. You also need to factor in is how your soundcard/DAC handles the output whether you output via SPDIF or use a USB DAC. If you are able to control the volume from your PC through your high end system, then something is not right

[filler56789]

@aquamarinebeach: you "slightly misunderstood" :) what I wrote — I didn't mean that already-existing build by [ReX], I meant the "non-existent" build based on that "obsolete" DirectX SDK.

Anyway, I've just run a brief, *totally subjective/unscientific* comparison between the available MpcMpaDec 1.0.0.0 and "the other decoders" :) To be honest, I did notice some differences in the output quality, BUT the differences were smaller (and therefore less significant) than I was expecting. :o Yes, the ancient MpcMpa decoder seems to apply less intensity on the lower frequencies, and the vocals have less distortions or artifacts with it. :confused: But then again, I am no audiophile, I am an old and impatient man already :) , and I do prefer discarding some poorly-encoded MP3s to looking for a "superior" MP3 decoder. :sly:

Alright then, these were my 2 coins. :)

[aquamarinebeach]

bjd - I have 2 sound setups:

A) The one I am testing is my (computer room setup) - i.e. my M-Audio Revolution 7.1 sound card converts stereo sound into 5.1 analog sound using SRS Circle Surround II that's built right into its drivers. The Soundcard drivers themselves take care of bass redirection / crossover to a Klipsch Promedia Ultra 5.1 Multimedia Speaker System.
This is why the volume is being controlled from my PC - as in this case my sound card is my receiver and the Klipsch speaker system has its in-built amplifier.

B) The other sound setup (my living room) uses the Digital Spdif port of the M-Audio Revolution 7.1 sound card. Sound is sent to a Yamaha RX-V1900 7.1 Channel Home Theater Receiver. The receiver uses SRS Circle Surround II (yes I love that technology) to output the sound to a Boston Acoustics A 2310 HTS 5.1 Home Theater Speaker Package.

And the reason why I started this thread and am so impressed with the internal mp3 decoder of MPC 6.4.8.6 is because it makes (Sound Setup A) sound very close to (Sound Setup B)

Regarding WASPI, ASIO, please take a look at:

http://wiki.hydrogenaudio.org/index.php?title=Bypassing_Windows_Mixer

I have tried ASIO in the past and switched back to Directsound - because I like the sound / compatibility (i.e being able to play sound from more than 1 stream at the same time.....). Also, I do not find that WASPI / ASIO gives me better sound. In fact, I find that DirectX gives me better bass response.

Instead of bypassing the Windows Mixer all along, I just just have whatever sound devices I am not using ON MUTE

[aquamarinebeach]

filler56789 - Ah okay, did not know you knew about the [Rex] decoders - as it was in a different thread

http://forum.doom9.org/showthread.php?t=120465&page=757

Regarding the difference in sound quality between MPADec 1.0.0.0 and the other ones - for me it's more than subtle - but then may be it's because I am listening to them on 5.1 systems. The sound stage is therefore bigger. I am sure if I were to listen to them on stereo speakers or headphones, then my evaluation would be just like yours.

[bjd]

@aquamarinebeach thanks for the info on your set up. You have quite a lot of post processing going on with the audio which is bound to have an effect on the overall sound and clearly the decoder you are after seems to compliment this.

This is not something I would do personally and have never been impressed with the results when trying to create 5.1 from 2.0 sources. Keeping the signal as pure as possible from LAV audio to speaker (bit perfect SPDIF -> DAC -> Stepped Attenuator -> Power Amp -> Speaker) works best for me. Good luck in your pursuit.

[aquamarinebeach]

Thanks bjd: So I take it you use some real heavy duty floor standing stereo speakers...Nice sound card by the way....heard really good things about the Meridian

[bjd]

Actually have a pair of Quad 11Ls on heavy stands for the front stereo channels powered by a Quad 405mk3 with a customer built pure silver wired passive pre-amp acting as a input switch/volume control. 2 channel playback is fed from the PC spdif to a Quad 99CDP which acts as a 24bit DAC.

For 5.1 playback i use the Meridian as a pre-amp and output PCM straight to two high quality power amps (rear,front,centre,passivesub) plus the front to the front channel preamp/poweramp (no theatre receiver). Swapped out op-amps in the sound card for 8 x BurrBrown OPA627AU - gives a very clean surround sound. The 2G version of the card is reported to be even better through the use of improved caps on the board.