Slashdot Mirror
Linux Based Stereo Components
davidu writes "MP3.com has a really cool interview with Joseph Mesterhazy, the creator of the LCDAT Linux-based MP3 player. If you haven't seen it, it is extremely cool. The interview also talks about how open standards make projects like this easier. This is is one of the first, not vapor, stereo quality, MP3 players out there, and it runs Linux!
" I wonder how many of us have duct taped perl scripts together
for our MP3 playing.
Hopefully this can come of some use to people who know little about mp3s and CDs. I forget the exact number, but audio on CDs takes up somewhere around 1400 kilobytes per second. The majority of mp3s don't even reach 256 kilobytes per second. The reason that mp3s can sound so close to CD audio with the savings in size is through the masking of frequencies.
The phenomenon is almost equivilant to the difference between newsprint advertisements and a painting (not in quality, but in terms of perception). With the newsprint, the picture is actually made up of tiny dots which, when viewed from afar, blend together to make a seamless looking picture, even though there are gaps if you look up close. Mp3s use the same principle, only with sound. Say you have a sound with a frequency of 80hertz at 60decibels; because of how our hearing works, that sound being played will mask out sounds around it. So for instance, if you had that sound playing, and suddenly you played a frequency of 75hertz at 30decibels, there would be no audible difference (note: these are numbers I'm making up to illustrate the general idea. I don't know the actual mathematics of it).
What mp3s do is allocate a certain amount of space (128kilobytes/sec or 256kilobytes/sec), break down a second of music into a number of frames (usually 44.1) and then for each frame looks at what the frequencies are, and gradually and more agressively gets rid of the sound that is most likely to next be masked.
The problem that can arise with this fixed bitrate encoding is that the encoder is not making an overall judgement based on what the quality of each frame should be, it merely knows how much it MUST cut out of each second of music. The solution to this is VBR or variable bitrate encoding, which has not been very widely accepted as far as I've seen. The difference with this style of ecoding is that you specify to the encoding program a quality level to set the song, and then it will go through each frame and cut only enough sound to fit your preference for sound quality, not some arbitrary number.
Anyhow, I think that's enough of me rambling. Hopefully someone found that at least slightly informative and interesting.
Paelon
I'm putting this in the main thread 'cause the MD article I read (posted by Foxman98) was buried a level.
:). Even the $4000 stereo I put in my car didn't sound as good as a properly set-up soud stage in a home room. There's something about a larger room that just gives more ambiance.
I used to work at a high-end stereo shop, and we sold everything from Sony cheapo stuff to Sony ES to Denon and up. I got the chance to play around with an MD versus a CD. After some rigging, I was able to switch to and from the MD and CD, and I had the exact same music playing simultaneously. On cheaper speakers ($300/pr) it wasn't real easy to tell the difference, however on more expensive speakers ($1800/pr) the difference was night and day. Compared to the CD, the MD was a little shrill, and the base was a little "boomier". All in all, the MD would probably sound better in a car than a CD, simply because a car stereo usually doesn't reproduce quality highs and lows (mids are good tho
Anywaze, on to MP3. From what I read about the compression technologies, it seems like the MD and MP3 algorithims are similar, but MP3 is designed to run at 128K/s to 256K/s data streams, where MD's are about 512K/s. (Correct me if I'm wrong, it's been a few years) I'd bet that an MD compared to an MP3 in a car would sound identical, but on a quality home system the difference would be night and day. Now say we got a better MP3 player (XAudio extensions anyone?) that supported a mic-feedback EQ so the system could be tuned, it'd probably sound damn good -- especially for the price!
Just my $.02.
LOAD "SIG",8,1
LOADING...
READY.
RUN
Joseph is a member of mp3stereo@lists.gofast.net. If you're at all interested in issues about building your own stereo components, then you need to join the list. Send a message to mp3stereo-help@lists.gofast.net to learn how to sign up.
I think you're 100% right. MP3's are good when you've got mediocre equipment. If you've got good ears and high end equipment, they'll drive you nuts. I listen to them all the time at work on my crappy Polk speakers that came with this PC.
:) Lot of people don't understand that LOTS of people CAN hear the difference even at 256k.
At home? They're tolerable, but no comparison really to the quality from a CD played in my DVD player (which outputs digitally to the receiver).
In my car? Not a chance. If I'm driving on the highway or something, I won't notice the quality as much, but on side streets there's not enough road noise to mask it. Drives me crazy.
Try saying that on rec.audio.car though and see how quickly you get flamed.
This article seems to imply the empeg unit doesn't exist. Strange that mine is shipping as I speak.
/dev/null'ed.
I submitted an article last week about how they're shipping unit now, but it was
The empeg car player is *not* vapourware.
Hi, I am the author of this project, and would like to share my opinion about this whole MP3 audio-quality issue that a lot of people have brought up:
In short, for me it isn't one.
In the interview I stated I built this machine to replace my *analog* equipment. I don't think anyone can argue that a 256kb/s MP3 (on a digital output) sounds worse than a cassette tape. Perhaps if you bought a $1000 tape deck, but if this is the case I am sure you could care less about MP3.
Yes, if you listen for it, you can hear the compression of the audio. Especially in the rear channels if you place your receiver in a surround mode. If this distracts you to the point you cannot listen to the audio, fine. Don't use MP3. I encoded most of my MP3 at 160kb/s, and my classical music and ambient music at 256kb/s. As far as I am concerned it sounds *close* enough to a CD, that I don't mind if there is a little swishy-ness in the cymbals. The advantages of having all my music at at the touch of a button far outweigh this.
How many people do you know record TV shows at SLP speed on their VCR? Probably a lot. For them the advantage of being able to record 6 hours of video on a 2 hour (SP) tape makes up for the loss in video/audio quality.
It is all a matter of personal preference.
Ah yes, but if your speakers are vifa components driven by very expensive LEAP designed crossovers using frequency response charts produced from the specific drivers you're using in custom built 1" thick MDF enclosures that together produce the auditory quality of $20,000 home speakers, you can shut them up pretty quick. :) My only complaint is I'm running a MT configuration and the tweeter is a bit too off-axis for my taste, and I haven't gotten around to fixing it.
:)
When car audio is good, its *good*. But most people who think they've got something good have never heard a $20k set of speakers, much less a $250k set of speakers.
You may want to take a look at the Hoontech 4D Wave NX card, which uses the trident chipset of the same name. Trident has released a GPL'd driver, so it should work "out of the box" with ALSA ( Advanced Linux Sound Architecture). The Hoontech card has S/PDIF out (optical & coaxial). You can buy it from http://www.audiencedp.com for $47 plus $5 S&H (I did). You can also try to buy direct from Hoontech, but I am not sure how they handle the oversea shippings (unless you live in Korea :).