Cowon iAudio 7 (fw 1.17) fails on Vorbis, "Electrical" noise Options

[Squeller]

Hi,

received my iAudio 7 today and updated to 1.17. It fails on Vorbis: I get some kind of "electrical hiss" noise in the background, can't descibe in english. Really easily audible. Like a jammer in the electrical grid. The level of distortion is about the same level as if you encode to mp3/cbr 64/80. --> CRAP.
It's on Vorbis encoded at q5, no matter which encoder (oggenc gt2, aotuv, lancer). It's not there at mp3/lame/v0.
Didn't explore Cowon forums yet, maybe someone here knows the problem.

Anyone else?

Edit: I think this thread at HA addresses the issue. So is there any kind of firmware hack in order to solve the problem? I've sent a message to Cowon. BTW I think it's a shame. At the current state I'd say the iaudio7 does not support Vorbis.

EDIT (2008-11-06): I've uploaded some files, I wasn't very accurate at their length which does not play a role here. No effects on the Cowon side.

The original files:
http://www.fileden.com/files/2007/9/21/144..._orig_flac.flac
http://www.fileden.com/files/2007/9/21/144...orig_ogg_q6.ogg

The Cowon iAudio 7 output (note it has some extra white noise, but you get the idea)
http://www.fileden.com/files/2007/9/21/144...7-out_flac.flac
http://www.fileden.com/files/2007/9/21/144...a7-out_ogg.flac

This post has been edited by Squeller: Nov 6 2008, 09:42

[le_canz]

Hi,

I got the same problem with my iAudio U3, I hear a light noise in the background, like electrical parasite.

This is annoying indeed. I've tried many firmwares ; always the same.

I solved this issue by using Lame... (don't misunderstand the smiley, I love Lame ; what makes me feel sad is this bugged Vorbis support)

If you really want vorbis, stay away from iAudio. It's a shame because the player is extremely good... just this f***ing bug.

Sorry for my english

This post has been edited by le_canz: Nov 4 2008, 13:11

[Squeller]

Yes. It's not even a light noise, it's disturbing in silent parts of music, especially when you listen to music with more dynamics (Classical for instance).
In fact, this player lacks decent support for low bitrate content like for audiobooks etc. because of the failure of Vorbis-playback.

This post has been edited by Squeller: Nov 4 2008, 18:13

[BoneJazz]

Yes. It's not even a light noise, it's disturbing in silent parts of music, especially when you listen to music with more dynamics (Classical for instance).
In fact, this player lacks decent support for low bitrate content like for audiobooks etc. because of the failure of Vorbis-playback.

Hi Squeller

Try this forum: http://iaudiophile.net/forums/index.php. Maybe you can find a solution.

On a side note I have a D2 and I'm not experiencing any problem with ogg files.

[le_canz]

Yes. It's not even a light noise, it's disturbing in silent parts of music, especially when you listen to music with more dynamics (Classical for instance).

Exactly. Seems we have the same problem, with two different players.

[HotshotGG]

Edit: I think this thread at HA addresses the issue. So is there any kind of firmware hack in order to solve the problem? I've sent a message to Cowon. BTW I think it's a shame. At the current state I'd say the iaudio7 does not support Vorbis

The noise is being introduced by Tremor the integerized Vorbis decoder which is used on many embedded systems and DAP for faster decoding. Somebody in that other thread you posted to mentioned a global variable that is present. Appearently it has something to do with low accuracy decoding.

http://osdir.com/ml/games.devel.scummvm/20...6/msg00169.html

"* Tremor has also been upgraded with a low memory mode and a low
accuracy mode. The low memory mode, I haven't looked at in detail.
The low accuracy mode (which saves 50K of lib size due to reduced
tables) reduces the quality we get out. Supposedly it saves 15% of
CPU."

I think that's the problem right there or at least the noise you might be hearing could be roundoff noise due to the integerized nature of the decoder (I am sure there is better term for it in engineering). If my memory serves my correctly I think those are IMDCT tables or something of that nature that they are referring to. I am not exactly certain, but last time I looked at the source code there were a lot of integerized MDCT optimizations that were done that's the big difference between the main branch and Tremor. The regular MDCT branch routines use floating point values to perform calculations.

[Squeller]

Try this forum: http://iaudiophile.net/forums/index.php. Maybe you can find a solution.

No solution there, maybe it could be resolved by a firmwre file hack, but the firmware is checksummed. I'll see if WMA could be a solution or if I'll return the device because I don't think Cowon will resolve the issue.

[Rotareneg]

Yeah, I noticed this very soon after getting my iAudio 7. My "solution" was just to switch to LAME -V4 MP3s.

There's a port of Rock Box for the i7 and D2 that's being held back primary by the flash driver (or lack thereof.)

[Squeller]

I've totally overseen there's a pretty unknown Codec called "Windows Media Audio" I'm thinking about encoding my stuff to mp3 at higher bitrates (simply because the conversion in fb2k runs without intermediate file) and wma for lower bitrates. Hopefully wma pro. Conversion to vbr wma with fb2k runs fine except for the intermediate wav file - no progress indicator then.
Haven't tested if the Cowon plays back wma pro - I really hope so. Then it's a probably a very good alternative especially in the low bitrate area. A proprietary but modern codec - why not use it?.
Also the voice-codec is interesting for me. But don't know yet if the ia7 plays it ...

This post has been edited by Squeller: Nov 5 2008, 09:03

[vitos]

I wouldn't bet on that - WMA Pro is a completely different format and I haven't heard about any hardware player supporting it except Zune.

Hmm... Just downloaded iAudio 7 manual, and it doesn't clarify what kinds of WMA formats are supported.

[HotshotGG]

No solution there, maybe it could be resolved by a firmwre file hack, but the firmware is checksummed. I'll see if WMA could be a solution or if I'll return the device because I don't think Cowon will resolve the issue.

It's not so much firmware related as much as it has to do with Tremor. I think you are right though if you want somebody to blame you can blame Cowon for using the low accuracy branch.

[Squeller]

Quick test: It does not play WMA pro files, it just skips those.

This post has been edited by Squeller: Nov 5 2008, 14:40

[vitos]

BTW. Could you check if it plays AAC files..? Many Internet sources tell it does, but manual doesn't confirm that...

[Squeller]

No, it doesn't (Nero encoded, LC AAC). It doesn't display .mp4/m4a, if you rename it to e.g. mp3 it skips playback.

I got the expected answer of Cowon Germany support: "Please use our Mediacenter software, as for 3rd party software we can not guarantee ... blah".

I'll document the issue at appropriate places (more than once I read "hardly audible") and return the device.

[vitos]

Could you try renaming the file to .AAC extension..? Supposedly that's what makes AAC files play on Cowon D2, and its' specifications don't advertise this fact. I wonder if it is the case with iAudio 7...

[Squeller]

No displaying of .aac either.

[vitos]

Thanks, Squeller.
So no AAC and crippled OGG support... These are serious shortcomings in my opinion, considering price of this player. Even my dying iAudio U2 plays OGG flawlessly... iAudio 7 won't be its' successor however.

[BoneJazz]

Another suggestion (it works for me on my D2):
Try to preprocess your files with lossyWAV and then encode them with FLAC --portable preset or, for a low bitrate --q0 (see the wiki for info).
Last chance: I know that's not an elegant solution but maybe a firmware downgrade could works.

[Mike Giacomelli]

Edit: I think this thread at HA addresses the issue. So is there any kind of firmware hack in order to solve the problem? I've sent a message to Cowon. BTW I think it's a shame. At the current state I'd say the iaudio7 does not support Vorbis

The noise is being introduced by Tremor the integerized Vorbis decoder which is used on many embedded systems and DAP for faster decoding. Somebody in that other thread you posted to mentioned a global variable that is present. Appearently it has something to do with low accuracy decoding.

http://osdir.com/ml/games.devel.scummvm/20...6/msg00169.html

"* Tremor has also been upgraded with a low memory mode and a low
accuracy mode. The low memory mode, I haven't looked at in detail.
The low accuracy mode (which saves 50K of lib size due to reduced
tables) reduces the quality we get out. Supposedly it saves 15% of
CPU."

I think that's the problem right there or at least the noise you might be hearing could be roundoff noise due to the integerized nature of the decoder (I am sure there is better term for it in engineering). If my memory serves my correctly I think those are IMDCT tables or something of that nature that they are referring to. I am not exactly certain, but last time I looked at the source code there were a lot of integerized MDCT optimizations that were done that's the big difference between the main branch and Tremor. The regular MDCT branch routines use floating point values to perform calculations.

Has anyone confirmed that its the low accuracy mode thats the problem here (by comparing the D2 distortion to tremor's low accuracy mode on the PC)?

[Squeller]

Added files to the first post for you to listen yourself. At my first ogg listen I had some more serious artifacts, but can't remember which samples those were. The output is terrible enough anyway.

[bug80]

Even my dying iAudio U2 plays OGG flawlessly...

Wasn't this "OGG distortion" problem also present in older firmware versions of the D2?

[edit]Oh, I'm sorry, you said U2

Anyway, I believe the D2 is the only player where Cowon fixed the issue. Howover this does mean that it can be fixed[/edit].

This post has been edited by bug80: Nov 6 2008, 09:55

[ameyer17]

I think that's the problem right there or at least the noise you might be hearing could be roundoff noise due to the integerized nature of the decoder (I am sure there is better term for it in engineering).

Quantization error?
And it has nothing to do with integer decoder roundoff error.
My Rio Karma, various Rockboxed mp3 players (Sansa e200, Sansa c200, 2nd gen iPod mini with 32 GB CF) , and (presumably) Sansa Clips all use Tremor and don't have this issue

This post has been edited by greynol: Apr 2 2009, 20:48

[HotshotGG]

My Rio Karma, various Rockboxed mp3 players (Sansa e200, Sansa c200, 2nd gen iPod mini with 32 GB CF) , and (presumably) Sansa Clips all use Tremor and don't have this issue

The reason that I have for that is that they are probably using Tremor in a higher precision mode. If you noticed in the other thread I mentioned compiling a thread for numerous DAP that have this problem including the Cowon iAudio 7 and some other players, because the question is brought up so frequently.

[viktor]

someone who owns an i7 please check out the 1.18 firmware, as it states:

- Improved playback sound quality of OGG files.

source

can you confirm that the playback quality has really improved with this update?

This post has been edited by viktor: Jul 24 2009, 04:04

[cpchan]

can you confirm that the playback quality has really improved with this update?

Yes, the noise is gone.I have been happily using ver. 1.18 with Vorbis files since it came out. Reencoding my collection to my portable Vorbis quality (-q 4) was a pain, but well worth it.

This post has been edited by cpchan: Jul 24 2009, 06:41