Erm, if you could design, photoplot, produce and populate an empeg board (8-layer PCB, plus a 2-layer one for the frontboard) and the metalwork (we'll exclude the HDD) for $1k, you'd be doing amazingly well. You might have problems sourcing some of the chips though, some are in-car only products and aren't sold via distribution.
You're also overlooking the huge amount of effort that has gone into (and will continue to go into) the software. The empeg is totally software upgradeable: we only run at about 35% load playing MP3s, we've got loads of headroom for whatever comes along next. This isn't a disposable like a Rio, this product is built like a tank and is looking towards the future.
An existing SBC unit could be done - and if you want to, there are descriptions, pictures and source of my earlier project, the mp3mobile, here.
I've said before: there's no room for a CDrom (without losing the display) and a CDrom doesn't give the integrated music database concept room to move. You just get a big CD with *some* of your music on it. Yes, 12x as much as a normal CD, but it's still only *some* of your music. You have to swap disks. The empeg is all about having it ALL and having it NOW.
Well spotted. Ok, so we only have 8Mb total, but we use 4Mb of cache at the moment. We've got some advanced caching algorithms and we can give functionality which appears to require the drive spun up with it spun down, only surfacing every few minutes (at 128kbit).
The empeg is 1.0 DIN sized. If we added a laptop CD-ROM drive, we'd have to lose the display (or put a less cute one on there). The physical display (not the active area) extends to within about 2mm of the metal casing top & bottom.
The display itself is specced operating temperature down to -40degC. However, at that point the PSUs on the board might well decide they don't want to wake up:)
1. The bitstream is taken from the digital line-out on most home players and buffered then fed into a MP3 decoder. Great idea, but the bitstream is running at 12x the rate required for MP3s, and the digital outputs offer no way of pausing or stopping the cd player - what are you going to do, have 650Mb of ram buffer?!
2. Some sort of error recovery on resampled audio data (in the same way as a 56k modem decodes PCM from an original digital bitstream). Cons: less capacity but compatible with more players and a more expensive decoder.
Both ways you will blow your speakers if you play a mp3 CD on a standard CD player - some players will mute the DACs and give you digital output on tracks with certain formats, some won't.
This sort of thing was thought up at the time of video-cds: just hook a video CD decoder onto the digital out of an audio player. This was easier, as the data rate is at least the same, although the error correction isn't compatible.
What this system *won't* do is to let you play iso9660-recorded MP3 CDs. I have no idea how it'll work for playing individual tracks either - maybe you select the tracks for 15 minutes of mp3 play and it buffers the bitstream in ram? You won't get constant play as there's no method for the decoder to control the cd player unless it knows lots of IR codes and has some pretty advanced pattern matching to ensure that it doesn't get overlaps in the data it receives from the cd player after a pause/restart operation.
They'd have more luck making a home CD-player which played mp3s. Something which is well within the reach of the companies involved.
If you're making a USB device, you need a USB manufacturers ID. From memory, this is about $1000: if you're big you really need to join the forum. More $. Everything has a price, you just pay it in different ways. Intel "tax" lots of motherboards, even ones with non-intel CPUs, because of the chipset: noone appears to complain when the charge is contained in the cost of a component, eg the $2 licencing fee to Frauhofer for the MP3 patent in the 3507D decoder chip as used in the MPMAN and the Rio.
Grow up, $1 is a bargain for what is a high-end port on a machine. Any idea how much IEEE 1394 controller chips cost? Much more than $1...
...seeing as the Micronas Intermetall part (the 3507) is the one in the MPMAN and the Rio. But, it looks like MP3 decoding appearing in the sound chip itself (if it's like other ESS chips): the thing is, most STB's have enough CPU to do mp3 decoding in software anyway...
Yes, an ENCODER card would make a lot of people happy. Remember the $25 licence for use of the encoder to Frauhofer IIS though, and that to do MP3 encoding at a decent speed requires a fairly hefty DSP. Ideally, you want one with a IDE/SCSI bus, so it can rip & compress itself, to alleviate the CPU having to throw data around for no good reason (and hopefully, with a fast enough DSP, being able to rip and compress at say 10x).
I really can't see Compaq's marketing division selling a development board with no software: it's not their field. The PLEB project is working on a similar thing, a small SA1100 board for experimenters. For an Itsy in your car, get an empeg.
The Itsy wouldn't be cheap either. Even if they made them in the 1000-ish quantity I'd expect it'd be around $1000 each: still interested in a PDA without ethernet, software, which you have to re-flash every time you want to update the code at this price?
Slashdot Mirror
You're also overlooking the huge amount of effort that has gone into (and will continue to go into) the software. The empeg is totally software upgradeable: we only run at about 35% load playing MP3s, we've got loads of headroom for whatever comes along next. This isn't a disposable like a Rio, this product is built like a tank and is looking towards the future.
An existing SBC unit could be done - and if you want to, there are descriptions, pictures and source of my earlier project, the mp3mobile, here.
I've said before: there's no room for a CDrom (without losing the display) and a CDrom doesn't give the integrated music database concept room to move. You just get a big CD with *some* of your music on it. Yes, 12x as much as a normal CD, but it's still only *some* of your music. You have to swap disks. The empeg is all about having it ALL and having it NOW.
Hugo, empeg
However, it has irda, so you could feasibly download pics from your digital camera to the hdd in the unit via the front-panel IR :)
Hugo, empeg
Hugo, empeg
Hugo, empeg
Hugo
1. The bitstream is taken from the digital line-out on most home players and buffered then fed into a MP3 decoder. Great idea, but the bitstream is running at 12x the rate required for MP3s, and the digital outputs offer no way of pausing or stopping the cd player - what are you going to do, have 650Mb of ram buffer?!
2. Some sort of error recovery on resampled audio data (in the same way as a 56k modem decodes PCM from an original digital bitstream). Cons: less capacity but compatible with more players and a more expensive decoder.
Both ways you will blow your speakers if you play a mp3 CD on a standard CD player - some players will mute the DACs and give you digital output on tracks with certain formats, some won't.
This sort of thing was thought up at the time of video-cds: just hook a video CD decoder onto the digital out of an audio player. This was easier, as the data rate is at least the same, although the error correction isn't compatible.
What this system *won't* do is to let you play iso9660-recorded MP3 CDs. I have no idea how it'll work for playing individual tracks either - maybe you select the tracks for 15 minutes of mp3 play and it buffers the bitstream in ram? You won't get constant play as there's no method for the decoder to control the cd player unless it knows lots of IR codes and has some pretty advanced pattern matching to ensure that it doesn't get overlaps in the data it receives from the cd player after a pause/restart operation.
They'd have more luck making a home CD-player which played mp3s. Something which is well within the reach of the companies involved.
Grow up, $1 is a bargain for what is a high-end port on a machine. Any idea how much IEEE 1394 controller chips cost? Much more than $1...
...seeing as the Micronas Intermetall part (the 3507) is the one in the MPMAN and the Rio. But, it looks like MP3 decoding appearing in the sound chip itself (if it's like other ESS chips): the thing is, most STB's have enough CPU to do mp3 decoding in software anyway...
Yes, an ENCODER card would make a lot of people happy. Remember the $25 licence for use of the encoder to Frauhofer IIS though, and that to do MP3 encoding at a decent speed requires a fairly hefty DSP. Ideally, you want one with a IDE/SCSI bus, so it can rip & compress itself, to alleviate the CPU having to throw data around for no good reason (and hopefully, with a fast enough DSP, being able to rip and compress at say 10x).
a development board with no software: it's not their
field. The PLEB project is working on a similar thing,
a small SA1100 board for experimenters. For an Itsy in
your car, get an empeg.
The Itsy wouldn't be cheap either. Even if they made
them in the 1000-ish quantity I'd expect it'd be around
$1000 each: still interested in a PDA without ethernet,
software, which you have to re-flash every time you want
to update the code at this price?