Okay this one's weird. Real weird. I'm slowly filling up my 18GB Empeg with MP3's as I rip them. So, I'm syncing about once a day. In addition to the frequent disk integrity checks from Emplode, I'm having the oddest problem. I first noticed it when I was demoing my Empeg to some buddies and a song was playing and in the middle it cut into a bunch of garbled nonsense, then went back to playing normally. I rewound to check if it was a minor sound glitch, but it was the same mess each time at that position in the song. So it sounded like a garbled MP3. However I've listened to the source MP3's and they're fine. So the possibilities as I see them are:

1. Errors over USB during sync (doesn't sound too probable does it?)
2. Corrupt data on the disk
3. Little gremlins inside my Empeg moving my data around

What's the deal? I haven't heard anyone else complaining about any problems like this. I've noticed it in about 10 songs. When I delete them and re-sync them they're fine. But I'm guessing this isn't a good thing, and I'd like to know what y'all think, especially the Empeg guys. I'm hesitant to report it as a bug until I can identify what the heck is going on under the hood.

---
MkII 080000554

I was about to say that the Empeg's MP3 decoder might be more sensitive to glitches in the MP3s than WinAmp, and that the files might really have some bad frames. But then you said this:

"When I delete them and re-sync them they're fine."

And that's really scary. The files really could be corrupt. The question is (as you surmised), were they corrupted by the USB transfer process, or is the Empeg's disk bad? Another possibility is a RAM problem on the Empeg player...

Whatever it is, you seem to have gone through a good set of logical deduction steps. The glitches in the MP3s simply shouldn't be there at all if the source files are OK. I will tell you that this has never happened to me on my Mk1 player. The only time my MP3s glitched was when my hard disk went bad, and there was a span of about 2 minutes between the time that I noticed a glitch and the time when the hard disk completely failed and wouldn't boot.

___________
Tony Fabris

Well, I think corrupt RAM is out of the question because it would be a sporadic problem and affect all MP3's. Certain ones are good, others are bad, and bad in the same spots. I don't know, I'll be doing a lot more syncing in the near future so I'll continue to report the situation. I hope I can squash this one though.

---
MkII 080000554

Another possibility is a RAM problem on the Empeg player...

I have finally diagnosed all the problems that I have been having to be RAM related. (Waiting for Rob to arrange warranty job... *poke* *poke*)

Strange pops and squeaks that don't exist in the original files are a symptoms that I have noticed too, although they are not generally reproducible between reboots.

I wonder what sort of error-detection is performed on the MP3 files as they are being transmitted to the Empeg. It might be worthwhile for emplode to calculate a checksum for each MP3 file that is uploaded, then ship both the file and the checksum to the Empeg. If the Empeg cannot reproduce the same checksum, then it knows that one or the other is corrupt.

Those checksums could also be stored on the Empeg itself as an extra integrity check... Every time the file is loaded, the checksum could be re-computed, and if it differs from the expected value an error event could be generated...


Michael Grant
12GB Green
080000266

Greetings!

What release of the software are you running, and are you running the developer image? You might want to try
setting up an FTP or HTTP server, or in some other way getting access to the file as saved on the empeg. Then
you can verify that the file is corrupt on the empeg's disk.

Another thought - a while back on this board I saw something with a fingerprint mechanism for MP3s. It might
be possible to get a fingerprint of the source and "on empeg" file, and compare. This could be another way to
validate if the on disk file is corrupt.

Lastly, you might want to try doing a sync of the corrupted files over ethernet, and see if you still see problems.
That would eliminate the USB upload as the cause, if you get corruption on the new batch.

Paul G.
Q# 15189
Status: In queue for a green 36GB Mk2

I had exactly what you describe in the very beginning (Release 0.5 about a year or so ago) when I was synching over serial, ie. good source files that played cleanly on the PC and then glitched/fizzed/popped/splatted/garbled on playback on the empeg. The clue was the repeatability on specific tracks, not randomness.

What was happening was that during the download of sequential blocks of data, the odd block was dropped, and never re-tried. This left "holes" in the downloads, which the decoder interpreted as a "Splat" (technical term)

Mike fixed it with several small iterations of the download protocol (which were also applied to USB) which made it more robust.

I would think it would be worth the effort to report it - but it will be difficult to track as a bug. Your "delete and download again" workaround looks to be the only current solution.

Mail it to bugs.

One of the few remaining Mk1 owners... #00015

Hmm... I want to see if it continues to happen. Since I'm running player 1.0 (developer image) and nobody else is having this problem, I'm hesitant to scream "fire" just yet. I'm going to do some more sleuth work to see if I can diagnose it as a hard drive or a transmission problem. Dunno how I'll do that, but that's the fun part of being a hacker.

---
MkII 080000554

In my limited use of the empeg (3 weeks), I have seen this a couple time. Once was missing data from the middle of the song, once was additional 5 sec at the end (specifically repeated the last 5 sec at half volume). stange stuff??? I have win 98 using USB, Mark 2 version 1.00. After I deleted and resynced the files were fine.

If resyncing wasn't so hard, this wouldn't be as big a deal...


-Doug
(Mk2-12G-Blue)

I have also seen this! Although I will admit that I just assumed it was my MP3's, however now I'm not so sure - I will have to listen in on a pair of headphones now :) Actually, I was playing the thing quite loud the other day and it 'screeched', I actually thought a very large rock had hit the car!! Then my girlfriend said it was the empeg.

I also get seemingly random reboots during the middle of a song. Entirely unreproducable on a song-by-song basis :(. I just figured it was a dodgy ver 1.0 bug or something.

Maybe there's more to this than at first meets the eye. I shall listen to a track that I know exhibited the problem to see if it's clean.

M

----
MKII #080000528 - 40Gb Blue

The easy way to test this is this: when you see the problem (the problems we've seen are only when the unit is warm - we've modified the soak test to contintually test ram during the soak now) start a terminal and power cycle the empeg, pressing Ctrl-T when you see the first message from the empeg.

This enters the onboard test code, which will test various things (including the ram). If you see any faults, email the log to [email protected]

Hugo

That Ctrl-T RAM test only works in the developer build, right?

___________
Tony Fabris

No, Ctrl-T is in the initial boot firmware and works even if you actually remove all the ram from the unit (though obviously it'll say the ram doesn't work ;) ).

It's only in mk2's though. We're not aware of any mk1's ever having memory problems - seems like Samsung ram is better than Siemens stuff...

Hugo

(Hugo beat me to it, so I deleted my reply).

Edited by Roger on 27/9/00 04:42 PM.

I can vouch for that statement - they keep the best for their mobile phones

One of the few remaining Mk1 owners... #00015