Open Source Firmware For iRiver players, development by Rockbox team starts Options

[rjamorim]

Interesting quote of Rbx iRiverport Wiki:

AAC (MP4)   Code fully-working in simulator but not yet on iRiver, not yet checked in.

Amazing indeed. I wasn't expecting AAC/MP4 support so soon.

Cheers!

[unfortunateson]

also seems as though the devs have optimized the Musepack code enough so it now can playback in realtime on the the Iriver, just gotta wait for them to release it

This post has been edited by unfortunateson: Oct 12 2005, 22:54

[RotAtoR]

also seems as though the devs have optimized the Musepack code enough so it now can playback in realtime on the the Iriver, just gotta wait for them to release it 

Very good news indeed. Unfortunately they also said it's somewhat of a hack and only a temporary fix until the mpc decoder can be rewritten to better suit the platform. It sounds like playback will probably be just barely realtime for most files.

This post has been edited by RotAtoR: Oct 12 2005, 23:28

[de Mon]

also seems as though the devs have optimized the Musepack code enough so it now can playback in realtime on the the Iriver, just gotta wait for them to release it 

Very good news indeed. Unfortunately they also said it's somewhat of a hack and only a temporary fix until the mpc decoder can be rewritten to better suit the platform. It sounds like playback will probably be just barely realtime for most files.

Strange... Vorbis is easy playable... As I know MPC requires less CPU time than Vorbis

[rjamorim]

Strange... Vorbis is easy playable... As I know MPC requires less CPU time than Vorbis 

The obvious conclusion is that Vorbis is much easier to optimize than MPC.

The other conclusion is that there are more developers interested in optimizing Tremor than ones interested in libmpcdec.

[unfortunateson]

Should musepack support be added under the H1x0 series in the wiki? It now plays in realtime (although there are still seeking bugs, unoptimized, not yet gapless). I don't know what format I should transcode my Wavpack files in when I run out of space - there should be a formal test of vorbis vs mpc at standard (160-190) rates on rock music...

[Farpenoodle]

It's been done with classical music at least.

But yes, I would be interested to see how the two perform on harder stuff.

[rjamorim]

Should musepack support be added under the H1x0 series in the wiki?

If the author himself didn't add it, it probably means he doesn't consider it ready enough yet.

[Lear]

The obvious conclusion is that Vorbis is much easier to optimize than MPC.

I've only looked a little at it myself, but the impression I got was that libmusepack does lots of 32 by 32 bit multiplies, where the location of the "interesting" 32 bits in the 64 bit result can vary a great deal. For tremor (the Vorbis decoder), the "interesting" result is always in the top 32 bits. For the CPU in the h1x0 players, this is much easier to process.

[M]

... now if only they would consider trying to write support for the iMP- series!

- M.

[Sebastian Mares]

... now if only they would consider trying to write support for the iMP- series!

    - M.

True that!

[rjamorim]

... now if only they would consider trying to write support for the iMP- series!

    - M.

True that!

Donate some iMPs to the development team and I'm sure they would look at that

[AtaqueEG]

Is this firmware being developed for any CURRENT player in the market?

I would love to use it, but I can't find any compatible player! And I have given up on eBay. It is full of scammers!

[rjamorim]

Is this firmware being developed for any CURRENT player in the market?

Sadly, no.

[SebastianG]

(on mpc playback)

Strange... Vorbis is easy playable... As I know MPC requires less CPU time than Vorbis 

This is really odd. According to my experience with codec stuff the inverse filterbank is one of those quite CPU-power-demanding building blocks. If MPC's filterbank implmentation is the one to blame (fast on PCs but bad to port) they could just reuse the 32 band PQF filterbank code from their mp3 decoder since it's actually the same.

But then again... The current musepack stream format is kinda horrible.


Sebi

[HisInfernalMajes...]

Is this firmware being developed for any CURRENT player in the market?

Sadly, no.

Actually, they [might start] developing it for the iAudio X5. Here is the offical site for it on the RockBox website and here is a discussion about it on a iAudio fan page.

This post has been edited by HisInfernalMajesty: Oct 13 2005, 22:00

[jaybeee]

"Hi all,

As you probably know, the current FLAC decoder in Rockbox is based on the official libFLAC decoder, and this only just runs in real-time on the H1x0.  The nature of libFLAC has made it very hard to optimise.

I spent some time yesterday with another FLAC decoder - the one written as part of the ffmpeg project.  I ported it to Rockbox, with the result that it already (without any real optimisations) running about twice as fast as our original libFLAC decoder.  The CPU is almost constantly at 45MHz, with very little time spent "boosted" to 124MHz.
...

and

"... This new decoder (without seeking) is now committed to CVS and is therefore the standard FLAC decoder in Rockbox.

I'll add seeking support ASAP, but can't promise when.  But this gives others chance to work their optimisation magic and get it working even faster, or even for someone else to implement seeking.

Regards,

Dave.
..."

Good stuff eh.

[Tang]

Is this firmware being developed for any CURRENT player in the market?

Sadly, no.

Actually, they [might start] developing it for the iAudio X5. Here is the offical site for it on the RockBox website and here is a discussion about it on a iAudio fan page.

The firmware is being ported on iRiver H3xx... Linus is working on it, seems it progresses...

[rjamorim]

Is this firmware being developed for any CURRENT player in the market?

Sadly, no.

Actually, they [might start] developing it for the iAudio X5. Here is the offical site for it on the RockBox website and here is a discussion about it on a iAudio fan page.

The firmware is being ported on iRiver H3xx... Linus is working on it, seems it progresses...

Erm... that one is not being manufactured anymore either.

[Emanuel]

Erm... that one is not being manufactured anymore either.

Amazingly short lifespan on the H-series from iRiver. One can not stop wondering why. The margins or sales must have been comparatively low.

[jaybeee]

When: 12 Nov 04:00 Who: Andy
What: iRiver - Initial support for wav-recording in recording menu. Supports mic/line-in (and radio), monitor mode, time-splitting (and byte-splitting), pause/resume etc.
Things todo: Prerecording, peakmeter (should be simple), frequency other than 44.1 kHz, etc..

Some more great new from the Rockbox boys

[ilikedirtthe2nd]

The Rockbox Ipod wiki page has been updated to show the status of the driver implementation. A funtional bootloader is available.

[Emanuel]

What: iRiver - Initial support for wav-recording in recording menu. Supports mic/line-in (and radio)

Now that's good news. Recording radio is yet one more of the features that was never available on the original firmware.

EDIT: spelling

This post has been edited by Emanuel: Nov 14 2005, 14:41

[jaybeee]

It's still early days for it though. I tried the Line In recording last night and it froze and crashed. Had to recover the file via CHKDSK.
Then there appears to be glitches every 90secs - however, I need to check this with different sources before I say that's a problem.
More testers the better.

EDIT: the glitches / drop-outs were caused by the source (doh!), so Rockbox is looking good for that now. Still some issues with freezing of player when recording stops, but I'm sure it'll get sorted soon enough.

This post has been edited by jaybeee: Nov 14 2005, 23:25

[ilikedirtthe2nd]

Rockbox now with crossfeed:

add crossfeed dsp effect. Makes some music more enjoyable with headphones.