mpg123-devel — Mailing list for mpg123 developers

Re: [Mpg123-devel] How to adjust the sample rate

[tonyli <to...@av...>]

Dear Thomas,

It is still under ARM. My os is linux 2.6 and Freescale iMX21. I can 
play the same song and same platform normally by using splay and 
downsampling. But
-2 has no sound.

Thanks and best regards,

Tony.

Thomas Orgis wrote:

>Dear Tony,
>
>Is this still on your ARM machine? Could you give a bit of info, like the hardware you have, what OS - and finally, how you compiled mpg123?
>
>Am Wed, 23 Aug 2006 06:34:32 +0800
>schrieb tonyli <to...@av...>: 
>
>  
>
>>I ran mpg123 -2 jn.mp3 or mpg123 -r 22000 jn.mp3. But the result shows 
>>"No supported rate found". If I used the default the sample rate, the 
>>speed of song is very fast. So I must downsample. Pls advise it.
>>    
>>
>
>"No supported rate found" means that the audio output didn't allow to set a rate near the one you wanted. Also, the speed being too fast hints at a mismatch between track sample rate and the sample rate the output is set to.
>What is the audio rate of the file? Could you make it available to me for testing (few MB eMail is OK)? 
>Probably more important is your audio hardware and what rates are supported.
>
>We can sort this out with some more info about your audio hardware and the sound driver you used.
>
>
>Greetings,
>
>Thomas.
>  
>


-- 
Tony Li
Software Engineer
Avantwave Ltd.
Direct Line: (852) 2952 7106

Re: [Mpg123-devel] How to adjust the sample rate

[Thomas O. <tho...@or...>]

Dear Tony,

Is this still on your ARM machine? Could you give a bit of info, like the hardware you have, what OS - and finally, how you compiled mpg123?

Am Wed, 23 Aug 2006 06:34:32 +0800
schrieb tonyli <to...@av...>: 

> I ran mpg123 -2 jn.mp3 or mpg123 -r 22000 jn.mp3. But the result shows 
> "No supported rate found". If I used the default the sample rate, the 
> speed of song is very fast. So I must downsample. Pls advise it.

"No supported rate found" means that the audio output didn't allow to set a rate near the one you wanted. Also, the speed being too fast hints at a mismatch between track sample rate and the sample rate the output is set to.
What is the audio rate of the file? Could you make it available to me for testing (few MB eMail is OK)? 
Probably more important is your audio hardware and what rates are supported.

We can sort this out with some more info about your audio hardware and the sound driver you used.


Greetings,

Thomas.

[Mpg123-devel] How to adjust the sample rate

[tonyli <to...@av...>]

Dear Sir,

I ran mpg123 -2 jn.mp3 or mpg123 -r 22000 jn.mp3. But the result shows 
"No supported rate found". If I used the default the sample rate, the 
speed of song is very fast. So I must downsample. Pls advise it.

-- 
Tony Li
Software Engineer
Avantwave Ltd.
Direct Line: (852) 2952 7106

[Mpg123-devel] #mpg123 channel on FreeNode

[Nicholas J H. <nj...@ec...>]

Dear all,

Thomas and I have been talking on the #mpg123 channel on the FreeNode  
IRC network. We would like to invite any users/developers to join us  
there to talk about mpg123.

See http://freenode.net/ for information about how to connect to the  
FreeNode network.


nick.

[Mpg123-devel] mpg123-0.60-beta4 release

[Thomas O. <tho...@or...>]

The beta4 is on sourceforge,
check it out and mangle it on the test bench!

Alrighty then,

Thomas.

--=20
Thomas Orgis - Source Mage GNU/Linux Developer (http://www.sourcemage.org)
OrgisNetzOrganisation ---)=3D- http://orgis.org
GPG public key: http://thomas.orgis.org/public_key

Re: [Mpg123-devel] MPG123 under ARM

[Thomas O. <tho...@or...>]

Am Tue, 22 Aug 2006 05:38:05 +0800
schrieb tonyli <to...@av...>: 

> Dear Sir,
> 
> I just downloaded mpg123 .59 and .60 version, and compiled them under 
> ARM-arch. But it doesn't have the sound. Anyone know what is the 
> problem? It is urgent!!!

I was not aware of our current 0.60 tree building on ARM - we did not do any testing on ARM devices. 
What kind of system do you have?
You used the linux-arm legacy make target?

I would like to see official arm support again with configure. We'll have to change some bits where verbatim double/float variables crept in to support ARMs without floating point emulation...

But, since your build finished, you have at least a mpg123 binary than can decode files, yes?
So you should at least be able to do

mpg123 -w output.wav input.mp3

and then have the data in the wav file.
When that works with our current code, we can start sorting out the missing output. The linux-arm make target only knows OSS output, perhaps you just need one of the other output options.

Again, I'd be happy to have working ARM version of mpg123 - so, let's sort this out!

Greetings,

Thomas.

[Mpg123-devel] MPG123 under ARM

[tonyli <to...@av...>]

Dear Sir,

I just downloaded mpg123 .59 and .60 version, and compiled them under 
ARM-arch. But it doesn't have the sound. Anyone know what is the 
problem? It is urgent!!!

Tony.

-- 
Tony Li
Software Engineer
Avantwave Ltd.
Direct Line: (852) 2952 7106

[Mpg123-devel] PortAudio and SDL

[Nicholas J H. <nj...@ec...>]

Support for audio output via PortAudio and SDL is now in the  
Subversion repository.

I have also partially implemented ALSA support - but having a few  
problems with it...


nick.

Re: [Mpg123-devel] MP3_NEED_MORE

[Thomas O. <tho...@or...>]

Am Mon, 10 Jul 2006 22:48:09 -0400
schrieb Stephen A Chait <sc...@MI...>: 

> Hi,
> 
> I'm Steve Chait and I'm designing a hardware mp3 decoder  in Verilog. I've been
> using mpg123's mpglib as a reference to help me write my code. In order to


> generate output to test my Verilog with, I compiled the mpg123 source files
> into a project and ran it, using some mp3 test files as input. I simply
> modified the read function in main.c so that it read characters from the input
> file. Some of the time it worked, but quite a few times I got the MP3_NEED_MORE
> error (interface.c, line 152). What does this error mean? Am I doing something
> wrong by running mpglib by itself?

Well, looking at the code and main.c it plainly looks like this error means what it says: Need more data.
The loop I see in main.c seems to cover that:

	while(1) {
		len = read(0,buf,16384);
		if(len <= 0)
			break;
		ret = decodeMP3(&mp,buf,len,out,8192,&size);
		while(ret == MP3_OK) {
			write(1,out,size);
			ret = decodeMP3(&mp,NULL,0,out,8192,&size);
		}
	}

Read and decode, go on reading if decoding isn't possible (yet, anymore).
This "error" seems to be a usual thing in that procedure.

> 
> BTW, the mp3 test files that gave me errors were the sine waves from
> http://www.dr-lex.34sp.com/macast/mp3test.html and Morse code files from
> http://www.arrl.org/w1aw/morse.html.
> 

Strange, I wasn't able to decode any of them properly with my version of mpglib (svn://orgis.org/mpg123/trunk/mpglib).

I have to admit that we didn't deal with mpglib much yet. The nature of mpglib has been that it is starting point for other projects to include a mp3 decoder and then being tested and changed in these projects.
We want to update mpglib, too, and place it as a "real" library installed in the system with other programs (perhaps including mpg123 itself) linking to it.
But that is future, for now all the work is in mpg123. Our mpg123 is pretty much the old unchanged one included with version 0.59r.
Even with the one from Michael Hipp's development trunk I cannot play your files... what exact version (from where?) do you have, actually?
Mind showing your patch?

I don't see us working on mpglib before getting the first "new" mpg123 release out - without mpglib, I suppose.
But after that, we want to get mpglib up to date and marry it with mpg123 again. This will include code drift between all of mpglib, mpg123 and real-world mpglib / mpg123 code implementations of other projects (MPlayer has a quite lively mp3lib).
For now that means that you are probybly more expert on mpglib than I am:-/
But maybe there's someone on the list having more experience with it?

So then,

Thomas.

[Mpg123-devel] MP3_NEED_MORE

[Stephen A C. <sc...@MI...>]

Hi,

I'm Steve Chait and I'm designing a hardware mp3 decoder  in Verilog. I've been
using mpg123's mpglib as a reference to help me write my code. In order to
generate output to test my Verilog with, I compiled the mpg123 source files
into a project and ran it, using some mp3 test files as input. I simply
modified the read function in main.c so that it read characters from the input
file. Some of the time it worked, but quite a few times I got the MP3_NEED_MORE
error (interface.c, line 152). What does this error mean? Am I doing something
wrong by running mpglib by itself?

BTW, the mp3 test files that gave me errors were the sine waves from
http://www.dr-lex.34sp.com/macast/mp3test.html and Morse code files from
http://www.arrl.org/w1aw/morse.html.

Thanks,

Steve

Re: [Mpg123-devel] Non-initialized pointer passed to DCT functions?

[Stephen A C. <sc...@MI...>]

I stepped through the mpglib code using a few test mp3 files. The hybrid_block
field is all zeroes, initially. But every time the DCT12 or DCT36 function
executes, values are permanently stored in the hybrid_block array. So the point
of hybrid_block is that rawout1 and rawout2 are different every time the loop is
taken.

for (; sb<gr_info->maxb; sb+=2,tspnt+=2,rawout1+=36,rawout2+=36) {
       dct36(fsIn[sb],rawout1,rawout2,win[bt],tspnt);
       dct36(fsIn[sb+1],rawout1+18,rawout2+18,win1[bt],tspnt+1);
     }

Steve

Re: [Mpg123-devel] Non-initialized pointer passed to DCT functions?

[Thomas O. <th...@or...>]

Am Fri, 30 Jun 2006 12:42:59 -0400
schrieb Stephen A Chait <sc...@MI...>: 

> write my code. One problem I came across was in the III_hybrid function
> (mpglib/layer3.c, line 1818). A local pointer called block is set equal to
> gmp->hybrid_block, a global field that never seems to be initialized (a string

I must admit that I didn't deal with the mpglib code yet - the mpg123 project is getting again on track after transferring maintainership from the original author.
This looks strange, indeed - but I will have to analyse it without creator's insight as well:-/
So, my first closer look at this thing...
The first obvious idea from me is that this whole gmp is the data linkk between host app and library, see mpglib.h:

BOOL InitMP3(struct mpstr *mp);
int decodeMP3(struct mpstr *mp,char *inmemory,int inmemsize,
     char *outmemory,int outmemsize,int *done);
void ExitMP3(struct mpstr *mp);

There, it's mp (struct definition in same file).
Then, look at Interface.c:

/* Global mp .. it's a hack */
struct mpstr *gmp;

[further down, line 131]

int decodeMP3(struct mpstr *mp,char *in,int isize,char *out,
		int osize,int *done)
{
	int len;

	gmp = mp;

So, I guess we should just look at an app using mpglib to see how this data structure works. Lame does it...

Thanks for pointing it out.. After the first new mpg123 release, there should be a review of the mpglib anyway, also in terms of sharing more code.

Greetings,

Thomas.

-- 
Thomas Orgis - Source Mage GNU/Linux Developer (http://www.sourcemage.org)
OrgisNetzOrganisation ---)=- http://orgis.org
GPG public key: http://thomas.orgis.org/public_key

[Mpg123-devel] Non-initialized pointer passed to DCT functions?

[Stephen A C. <sc...@MI...>]

Hello,

I'm Steve Chait and I'm doing a project where I'm designing a hardware mp3
decoder  in Verilog. I've been using mpg123's mpglib as a reference to help me
write my code. One problem I came across was in the III_hybrid function
(mpglib/layer3.c, line 1818). A local pointer called block is set equal to
gmp->hybrid_block, a global field that never seems to be initialized (a string
search for hybrid_block in all of the source files came up empty except for
that one assignment in layer3.c:1822). Then, two more pointers (rawout1 and
rawout2) are set equal to portions of block and passed as arguments to dct36
and dct12. From the complicated arithmetic done using values in rawout1 and
rawout2, it is clear that they get initialized somewhere, and do not just
contain zeroes. But where, and with what data?

Thanks,

Steve

Re: [Mpg123-devel] mp3lib mmx optimizations: LSB rounding errors?

[Thomas O. <th...@or...>]

Hi again!

Am Thu, 5 Jan 2006 16:32:51 +0100
schrieb "ACS" <adr...@xs...>:=20

> I'd like to report LSB errors in the mmx optimized decoding routines.
> I stumbled onto those when I was hacking around in sources to add dithere=
d=20
> output.
>=20
> When decoded to pcm, file compare will show differences.

I finally took a second look at this issue. Yes, there are differences in d=
ecoded material, but these seem to be non-systematic and based on rounding =
(difference +/-1, very rarely +/-2). I'm no expert in the mmx assembler cod=
e but I'm not sure if one can fix this at all without killing the optimizat=
ion.

> Also decoded test results are downloadable at:
> http://dither123.dyndns.org (download section,  nodither.flac and=20
> nodither_mmx.flac)
> These special samples "zoom into the LSB" , so difference is hearable to=
=20
> all, not just those with golden ears.

Whoa! I see what you mean by zoom. It's astonishing that a dynamic range of=
 less than 10 distinct steps still yields some distinguishable sound.
In that range, the (supposed) rounding differences have a great impact, whi=
le on "normal" amplitude data that you can hear without applying a factor o=
f 600 I doubt that you can hear them. But since your whole approach of appl=
ying dithering somehow is grounded on the statement that rounding to intege=
rs _does_ matter for sound quality, I suppose you can year it;-)

I also tested your dithering code and I hear how it at least makes the high=
 noise level more smooth and continuous... on can argue that that is an imp=
rovementwhile, while for the test "hotel" mp3 _any_ variant sounds ugly nev=
ertheless;-)

I'd like to give the audience the possibility to judge by themselves, so I =
opted to include your patch as a build option (a make variable switches bet=
ween normal and dither decode_i586.s). For that I now want to make sure tha=
t you agree:

1. Your adition is included and part of mpg123 under the LGPL 2.1 license.
2. You get credit for that addition.
3. You give permission to me (or whoever takes over mpg123 maintainership i=
n future) to allow usage of this code under other terms than (L)GPL provide=
d that your credit remains intact; in _addition_ to (L)GPL - your code will=
 always remain freely available under (L)GPL.

In effect I request sort of a BSD-style license for your changes.

So, are you OK with this legalese?


Thomas.

--=20
Thomas Orgis - Source Mage GNU/Linux Developer (http://www.sourcemage.org)
OrgisNetzOrganisation ---)=3D- http://orgis.org
GPG public key: http://thomas.orgis.org/public_key

Re: [Mpg123-devel] Sonos use of mpglib

[Thomas O. <tho...@or...>]

Hi! 

Am Tue, 6 Jun 2006 23:55:16 -0700
schrieb Jesse W <je...@ne...>: 

> Just alerting you that there's a fork of mpglib available for download 

Thanks for the info, I'll have a look at it. Maybe it's even interesting;-)


Thomas.

[Mpg123-devel] Sonos use of mpglib

[Jesse W <je...@ne...>]

Just alerting you that there's a fork of mpglib available for download 
from a company that's using it in their products, Sonos (a home digital 
music distribution system), at 
http://www.sonos.com/support/downloads/GPL/

I am not on this list, so please CC me any replies.

Jesse Weinstein

[Mpg123-devel] Re: mpglib with dedicated project

[Camilo S. T. <cst...@uo...>]

Hi,

My name is Camilo and I am from Brazil.
I am developing a project in SourceForge.net called notSAD.
I hope this project turns into an useful and standard-like LPGL library 
for Audio Decoding.
But I would like to rely on thirdy-party libraries for doing audio 
decoding librarian.
notSAD would only define an interface for notSAD plugins.
I could depict notSAD role in a succesfull audio playback scenario like 
this:
- an application request to play an audio file;
- notSAD matches the best plugin (wrapper);
- the plugin calls the decoder;
- the decoder returns decoded data.

----------------------      ----------------      
-----------------------      ------------------------
| Client Application | ---> | notSAD       | ---> |notSAD mpglib Plugin 
| ---> | mpglib for example   |
----------------------      ----------------  |   
-----------------------      ------------------------
                                              |
                                              |   
-----------------------      ------------------------
                                              --> |notSAD Plugin 2      
| ---> | another LGPL plugin  |
                                                  
-----------------------      ------------------------

But I would not like to include or to develop mpglib inside the hosted 
notSAD project.
And I always find mpglib inside another project, but not as a separated 
project.
I think also a separated project for mpglib would help make it more robust.
I didn't use MAD library, because it is GPL, and it does not give 
freedom for
closed, mixed or transitional projects. It limits to open source code.
As mpglib - I hope the owner of mpg123 project is the owner of mpglib and
that this owner read this email - is LPGL, I would like to reference and 
use it
in my project, but it would be better if there was no duplicated code.
I would event suggest a project dedicated to mpglib only.

Camilo S. T.

Re: [Mpg123-devel] mpglib as a SourceForge.net project

[Thomas O. <tho...@or...>]

Hi there, Camilo!

> - an application request to play an audio file;
> - notSAD matches the best plugin (wrapper);
> - the plugin calls the decoder;
> - the decoder returns decoded data.

Hm. A classic problem... But you are already at solving this one;-)

> And I always find mpglib inside another project, but not as a separated 
> project.
> I think also a separated project for mpglib would help make it more robust.

> As mpglib - I hope the owner of mpg123 project is the owner of mpglib and
> that this owner read this email - is LPGL, I would like to reference and 
> use it

I'm in the process of taking maintainership for mpg123. Part of this process is making the license clear.
As you may have noticed, Nicholas created this sourceforge project with a current debianized mpg123 under GPL following the last statement of the original author. This last statement also repeated that mpglib is LGPL.

Currently, mpglib and mpg123 are separate things (mpglib being pulled out for the very purpose of being included in other projects) with separate licenses. Current intention is to change that in both respects.
It makes sense to blow all the neat mpg123 decoding stuff (including optimizations, equalizer - IMHO) into an mpglib as "real" library and have the mpg123 part specializing in the interface(s).
Kindof weird conslusion of this would be to support different audio formats in the future through using different libraries... in that case we can still adopt a name like any123;-) 

So, yes, we intend to make a distinct mpglib not necessarily to be hidden inside several projects - as a source for an "official" mpglib.dll that I guess you'd use since your project looks very Windows-ish... you are invited to help us with supporting that specific OS, btw. (as I code mainly under Linux/Unix and Nicholas is more on MacOS).

And yes, the decision for _all_ of mpg123 simply being LGPL is kindof made - I'm going to contact all contributors I can get to ask for admission and then go for a fresh release of the state-of-the-art mpg123 with some accumulated work that lived in form of patches for too long.
Following that, there is still some space for discussion about where to go exactly with the development. Making mpglib the decoding engine of the player is a settled thing (IMHO - anyone objecting?) - feel free to share further ideas!

On the organization (sourceforgization): I think the mpg123 project will serve the purpose for the forseeable future. We intend to provide files via sf.net and use the collaboration features (like this mailing list).
Home page and subversion repository will be independed of sf.net at the known locations (mpg123.de/org/..?)
Maybe there will be some spawned project in the future, but for now it's all mpg123.



Thomas.


PS: Nicholas: have you yet tried to integrate some stuff into svn - MacOS patch, autoconf (or something simpler doing the same task just as good)? Also: activating the sf.net bug tracker may be a good idea. We already have the occasional posting here that should not be forgotten (that assenbler bug in post-0.59s, this ppc-nas patch).

-- 
Thomas Orgis - GPG public key: http://thomas.orgis.org/public_key
OrgisNetzOrganisation ---)=- http://orgis.org

[Mpg123-devel] mpglib as a SourceForge.net project

[Camilo S. T. <cst...@uo...>]

Hi,

My name is Camilo and I am from Brazil.
I am developing a project in SourceForge.net called notSAD.
I hope this project turns into an useful and standard-like LPGL library 
for Audio Decoding.
But I would like to rely on thirdy-party libraries for doing audio 
decoding librarian.
notSAD would only define an interface for notSAD plugins.
I could depict notSAD role in a succesfull audio playback scenario like 
this:
- an application request to play an audio file;
- notSAD matches the best plugin (wrapper);
- the plugin calls the decoder;
- the decoder returns decoded data.

----------------------      ----------------      
-----------------------      ------------------------
| Client Application | ---> | notSAD       | ---> |notSAD mpglib Plugin 
| ---> | mpglib for example   |
----------------------      ----------------  |   
-----------------------      ------------------------
                                              |
                                              |   
-----------------------      ------------------------
                                              --> |notSAD Plugin 2      
| ---> | another LGPL plugin  |
                                                  
-----------------------      ------------------------

But I would not like to include or to develop mpglib inside the hosted 
notSAD project.
And I always find mpglib inside another project, but not as a separated 
project.
I think also a separated project for mpglib would help make it more robust.
I didn't use MAD library, because it is GPL, and it does not give 
freedom for
closed, mixed or transitional projects. It limits to open source code.
As mpglib - I hope the owner of mpg123 project is the owner of mpglib and
that this owner read this email - is LPGL, I would like to reference and 
use it
in my project, but it would be better if there was no duplicated code.
I would event suggest a project dedicated to mpglib only.

Camilo S. T.

Re: [Mpg123-devel] Patch for PPC NAS support

[Thomas O. <tho...@or...>]

Am Fri, 17 Feb 2006 11:15:54 -0500 (EST)
schrieb mp...@us...: 

> The attached file adds support for building a ppc binary with nas support.

Thanks; hopefully we'll get such stuff on via configure script in the future, though;-)

> This tool is great by the way. I currently use it to stream line-in on one computer to all my other machines with the following command:
> 
> sox -t ossdsp -w -s -r 44100 -c 2 /dev/dsp -t raw - | lame -x -m s - - | mpg123-nas -

Hm. Though I also think that mpg123 is a useful tool, but in this line it's just sound -> compress, decompress -> NAS (right?).
Somehow sth. like libaudiooss (http://tvilsom.org/projects/audiooss.html) to get the sox output more directly to NAS could avoid this rather unnecessary step of lossy compression...

-- 
Thomas Orgis - GPG public key: http://thomas.orgis.org/public_key
OrgisNetzOrganisation ---)=- http://orgis.org

[Mpg123-devel] Patch for PPC NAS support

[<mp...@us...>]

The attached file adds support for building a ppc binary with nas support.

This tool is great by the way. I currently use it to stream line-in on one computer to all my other machines with the following command:

sox -t ossdsp -w -s -r 44100 -c 2 /dev/dsp -t raw - | lame -x -m s - - | mpg123-nas -

--
Thanks,
Marc

Re: [Mpg123-devel] mp3lib mmx optimizations: LSB rounding errors?

[Thomas O. <tho...@or...>]

Ho!

> Great to see that the most CPU efficient mp3 player isn't dead!

It is not dead what eternally lies and in strange aeons even death may die... or so;-)

> I'd like to report LSB errors in the mmx optimized decoding routines.

Good to have something to do after getting on with the "new" mpg123 at all;-)
Updating the asm routines is a thing I'd like to do anyway - after learning a bit more assembly magic...
These optimizations are pretty much what mpg123 is about.

> I stumbled onto those when I was hacking around in sources to add dithered 
> output.

Hm. Maybe a good thing to have as an option in the mainline mpg123... I'll listen to the test files tomorrow.


Thomas

-- 
Thomas Orgis - GPG public key: http://thomas.orgis.org/public_key
OrgisNetzOrganisation ---)=- http://orgis.org

[Mpg123-devel] mp3lib mmx optimizations: LSB rounding errors?

[ACS <adr...@xs...>]

Great to see that the most CPU efficient mp3 player isn't dead!

I'd like to report LSB errors in the mmx optimized decoding routines.
I stumbled onto those when I was hacking around in sources to add dithered 
output.

When decoded to pcm, file compare will show differences.
Also decoded test results are downloadable at:
http://dither123.dyndns.org (download section,  nodither.flac and 
nodither_mmx.flac)
These special samples "zoom into the LSB" , so difference is hearable to 
all, not just those with golden ears.

Adrian