Slashdot Mirror

The only MP3 player that I know of which will work with a non-FAT filesystem is the iPod, which on the Mac can be formatted as HFS/HFS+. Unfortunately, I don't think this is really going to help you as a Linux user, as you'll probably need a Mac to reinitialize/reformat the iPod to use HFS/HFS+, may need to install extra software to be able to

The iPod looks like any other usb-storage device to a Linux system. In order to switch the iPod to HFS+ using a Linux system, you will need to compile support into your Kernel for "Apple's Extended HFS File System" and "Macintosh partition map support" which is available in both the 2.4 and 2.6 kernels. You will also need HFSUtils which are installed with Slackware (I don't know about other distros).

mount HFS/HFS+ partitions under Linux, and will need to run iTunes through some sort of emulator to manage your music. Ick.

Or you could use one of the open source iPod applications such as gtkpod, gnupod, mypod, or one of many others. I've only used GTKPod which has worked great for me.

BTW What's going on here? Decoder tests show that later versions of Winamp are among the best. Why do mad and QuickTime seem better.

I don't remember the numbers, but MP3 licensing costs are insignificant. It's like less than a dollar per decoder.

Vorbis-decoding can be done using only integers (FLAC too?), which must save some hardware costs.

MP3 decoding can also be done with only integers. Cheap players already do this, so doesn't save you anything.

So you'd be offering a player with no real advantages except a miniscule price decrease, and some major disadvantages (transcoding). As much as I love Vorbis, that wouldn't work.

Much more customizable audio output support for one, can push out 24bit/96Khz if your card supports it, better dithering at 16 bits, etc. etc.

There's a MAD plugin for Winamp that decodes to 24 bits. The waveOut output plugin is supposed to carry those bits to the soundcard without rounding down to 16 bits.

Foobar can do gapless mp3 playback which according to hydrogenaudio discussions I've followed, looks to be pretty hard. The short gaps between songs that are gapless on CD when my iPod plays 'em back is quite disturbing. For instance, the Meridian Art Ensemble's Prime Meridian has Peaches En Regalia flow right into Let's Make The Water Turn Black and it's gorgeous, but the segue is wrecked on the iPod.

Foobar 2000 pushing the envelope doesn't let Apple get away with "it's too hard."

The WinAmp MP3 decoder is really poor quality. There's a plugin to make it use the MAD MPEG decoder, which will give you much better sound. I don't know if the plugin works with WinAmp 5; perhaps someone can try it and report...

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

OK, so it's all very well that you can now use SACD with more accurate signal reproduction, or even DVD-A (isn't that a term used in porn movies? So I've heard) if you want better quality.

Whose ears are actually good enough to listen to 24-bit audio and tell the difference between that and 16-bit anyway? I have often heard it said that analogue transmission of audio is far worse than digital. I don't entirely agree with that, but supposing it's true - surely the cables between SACD player and amplifier, amplifier and speakers are going to withdraw a lot of the benefits of the more accurate signal?

Yes, we can only hear about 20-bit accuracy. The point of the additional accuracy is, therefore, questionable. The difference in quality it will make is miniscule. The LSB on 16-bit audio represents a variation of 0.0015% in the output signal. The LSB on 24-bit audio represents a variation of 0.000006% of the output signal. Can you hear that final bit? Does it make all the difference? Er, no.

Those who say that the MP3 format is too lossy for them might be interested to know that audiophiles can't actually hear the difference between 256kbps MP3 and the original CD recording. Those who think they need still more quality should perhaps check out the MAD plugin which has the ability to decode mp3s to 24-bit, recreating bits that weren't even there in order to improve quality.

As regards introducing watermarks as a kind of copy protection - well, that's just reducing the quality of the audio, which defeats the point of what you were trying to achieve in the first place.

MAD (MPEG Audio Decoder) is an example of someone writing cool GPL software and then successfully licensing it with a non-GPL license to many commerical products that do not want to share their code.

I like the way the MAD MPEG audio decoder is
licensed. It's GPL code, but the author is open about licensing it to commercial closed-source projects. I think this is a nice arrangement:

http://www.mars.org/home/rob/proj/mpeg/

jeff

I believe that applies in this case.

But should you not wish to get linux installed on a machine, look at these links (obtained by the google search on "Reading an HFS disk on Windows") :

HFS Utilities Homepage

Port of hfsutils to DOS

Good luck and have fun.

From a computational accuracy perspective, mpg123 is the only decoder you mentioned which produces output in full compliance with the MPEG standard.

-v

It can be overcome pretty well by booting from flash memory. That's what the empeg does; it can boot in about 8 seconds, maybe less.

MP3 decoding can be done with a SA-1100 CPU in anywhere from 24-43% of real time, depending on the decoder. In fact I am working on such a decoder.

-v

As the author of hfsutils, and someone who is actively investigating this matter, I can confirm that LinuxOne's LinuxMac product as currently distributed provides no real functionality except a GUI wrapper and the ability to create Macintosh HFS file systems by bundling hformat.

The fact that the GPL'd hformat is included without a conspicuous copyright notice, without a copy of the GPL, and without including or making an offer to obtain the source has understandably raised my concern. Since I own the copyright, I am in the process of pursuing a legal remedy.

If you've purchased a copy of LinuxMac, or if you have information or concerns about the way it is being distributed, I'd like to hear from you.

As the author of hfsutils, and someone who is actively investigating this matter, I can confirm that LinuxOne's LinuxMac product as currently distributed provides no real functionality except a GUI wrapper and the ability to create Macintosh HFS file systems by bundling hformat.

The fact that the GPL'd hformat is included without a conspicuous copyright notice, without a copy of the GPL, and without including or making an offer to obtain the source has understandably raised my concern. Since I own the copyright, I am in the process of pursuing a legal remedy.

If you've purchased a copy of LinuxMac, or if you have information or concerns about the way it is being distributed, I'd like to hear from you.

Lots of stuff works on the devices. tclsh was an easy port, but awfully large to just have around without a good reason. I ran the LambdaMOO server using JHCore. Robert Coie actually built a self-hosting gcc.

Of course, without any CF access yet, anything that won't fit on the 4M ramdisk is run over the 115kbps SLIP/PPP link on NFS...not the speediest disk you've ever used.

> and they are working on getting the GUI running (it seems to be already running on at least one of the developers' machines).

The demo works on my E-15 too. It's pretty much stock microwindows, with new touchpanel calibration code.