Official (non-beta) Ogg support?

[sdc395]

Hi All

I posted this question just a day or two before Phatnoise closed their forums so I thought I'd try my luck here instead.

I was asking about the status of the Ogg support project.  I'm not especially keen on the slightly hack-like (and <100% reliable) method of enabling gapless playback and was hoping to hear that a better decoding method had been developed.

Does anyone have inside info on Ogg support (or support for any other format, for that matter)?  Have Phatnoise forgotten about the Phatbox now?

Cheers
Simon

[sbingner]

I want to work on porting the official current oggplay to phatbox...  and make it read real ogg tags

[judb]

has there been any major improvements to the performance of the tremor decoder?  Tremor is the interger math version of the ogg decoder for ARM etc... AFAIK the mavrick cpu we have in the keg / phatbox was just too slow to really be effective at playback and thats why vince ended up giving up on ogg playback.

Vince if you are around could you chime in please?

Also I'd like to see us replace flac play with a vorbis tag and id3 tag capable player where either tag format would be supported. well, obviously the players are not responsible for tag info in the phatbox, but able to deal with either tag format and when phat4x comes up to speed it would hopefully support both.  PMM on the other hand would not be able to work with both.

[sbingner]

Also I'd like to see us replace flac play with a vorbis tag and id3 tag capable player where either tag format would be supported. well, obviously the players are not responsible for tag info in the phatbox, but able to deal with either tag format and when phat4x comes up to speed it would hopefully support both.  PMM on the other hand would not be able to work with both.

I was mostly thinking of flac actually, don't much care either way on ogg... it's not enough better than the available formats

[judb]

I really like ogg format.. I think it sounds much better than MP3 but thats just my opinion... i wish we had better ogg playback support (well I guess CPU speed is what we need right?)  anyhow updated flac would rock.

[dscline]

I thought they did officially support it now.  I've seen comments that have suggested that, anyway.  FWIW, I just got a phatbox last week.  While my experience is limited, I haven't had any major issues with OGG playback.  With the exception of gapless playback, it's played just fine without having to manually add oggplay or any of the tags in the ini.  Adding vorbis_eof_buf_secs=3
to the ini seems to have fixed the gapless playback problem I had, at least so far (just made that change tonight, so it's only briefly been tested).

I also don't understand the tag comments... my tags are working.

FWIW, I've been encoding using the latest aotuv beta 4, @ -q 4.5.  No drop-outs at all yet.

[judb]

the tag comments were directed at FLAC files.  PMM doesnt support reading the vorbis format tags from FLAC files, intstead it reqires ID3 tags in them.  most flac encoders insert vorbis style tags and so having an existing FLAC collection and adding it to the phatbox sucks.

[sdc395]

I have found Ogg playback to be <100% reliable on occassion.  The player seems to skip the last second or two of a track.  Also, with the hack required for gapless playback, the headunit's counter gets out-of-sync.

I was under the impression that improved streaming from disk to decoder would alleviate these minor annoyances.

So, has official Ogg development ceased?  What about FLAC (I have cross-track rewind problems) and AAC (poor support for the various sub-formats)?

[judb]

I am sure that coding happens on all those formats for ARM but I do believe that Phatnoise has stopped development / updates of the codecs for the V1.x phatboxes.

[Oaf]

I've been using Ogg Vorbis with the Beta3 and 4 tunings and it works well.

I second the comments on FLAC tags! Though FLAC is of limited use with a 20GB cart :-)

I _would_ like to get some more codecs working on the Phatbox. I've had some success with a couple of brief tests but things would be much easier if we had a couple of skeleton/demo programs to work from or a little bit of documentation! ;-)

[sdc395]

If official development really has ceased, Phatnoise should support a community effort by providing some documentation and the required support utilities (for signing, etc.)  However, I guess a development kit for the Phatbox's hardware would be expensive.

If I'd known this a few months back, I might not have bought my Keg.  As I stated above, Ogg, FLAC and AAC playback still need work, IMHO.

[judb]

Well, the problem with releasing what we need to do community devlopment is the very problem we are trying to overcome in the dms hack forum.

See, the whole system of the phatbox is locked down by the boot loader in the flash memory.  This checks the drive sigs and the file sigs.. if we could sign our own executables then we could in theory sign our own drives, there by bypassing phatnoise's copy protection scheme and costing them revenue.  They wont do it I'd bet money on it.  I wish they would though.

However, we are VERY close to getting the problem solved on our end thanks to the hard work of many very smart people and when we get that problem solved we can start running whatever we want on the box.

I just build the most recent libvorbis tremor codec for the phatbox arch but we would need to build a oggplay type exe which sends the data to the correct places for the phatbox to work.

As for FLAC, I can't find any specific builds for ARM but I havent looked at the main source tree, it may already include it.

[Vince]

Here's some clarifications:
The PhatBox Ogg Vorbis decoder (officially supported) uses the latest version of Tremor, which hasn't had any updates in a long time.  It can't handle the highest bitrate files, and to avoid skipping, can create a gap between tracks.  You can go to phatbox.sixpak.org to see some tweaks you can make to fix things, which will have some side-affects.  I'd suggest using vorbis_eof_buf_secs=3 for starters.  Unless there's been a Tremor update in the past few months I'm not aware of, the situation won't get any better.

As for FLAC files, the changes needed to support different tagging formats are NOT in the PhatBox software (firmware), but rather the PMM or desktop program which creates the playlists.  You can write your own program to create these, or join one of the other efforts on these forums to work on solutions for that.

[judb]

Good point vince.  I knew that the decoders used on the phatbox dont have anything to do with reading of tag information, I think that was stated elsewhere, perhaps another thread.  

I saw some mailing list activity this year about patches to the tremor tree for fixing memory leaks and whatnot so I was using that as a chance to toy with my crosscompiler setup in cygwin.  

thanks for the info!

[sdc395]

I hadn't realised Ogg playback on the PhatBox was now officially supported.  AFAICT, the PhatNoise website only lists MP3, WMA, WAV, FLAC, Audible and AAC.  Even Vince's site reads: "PhatBox Ogg/Vorbis support is in progress. Before it becomes an official feature, the testing versions will be available here."  I guess they're just out-of-date.

It's a shame Ogg can't be made to work more smoothly but I do appreciate that a real effort has been made to implement support as well as can be accomplished given the limitations of the hardware.

All I really want is support for gapless playback of a high-quality lossy format.  Would gapless playback of MP3 be possible if the player supported LAME's padding tags?

[judb]

Vince?  Care to chime in?

[dscline]

It's a shame Ogg can't be made to work more smoothly  ...  All I really want is support for gapless playback of a high-quality lossy format.

I still don't follow what the issues are.  I've encoded all my library to oggs, and since adding the buffer line in in inf, gapless playback has worked flawlessly for me.  People have mentioned that it's a "hack", and not reliable, but the only "issue" that I've experienced so far is the fact that the display on the head unit registers a track change a couple of seconds before it actually happens.  Hardly the end of the world.  Before I added the buffer line, the gaps seemed like less than one second to me, so I may even try setting it to two seconds (instead of three).

I'm not saying there aren't legitimate issues, but what ever they may be, I haven't noticed them.

[sdc395]

I'm not saying there aren't legitimate issues, but what ever they may be, I haven't noticed them.

The issues I've encountered are far from serious and I'm not complaining.  I just wanted to know if I could expect an update to the firmware that would iron out the few remaining problems.

I could be wrong (I confess I've not taken the time to check) but I'm sure the headunit counter becomes increasingly out-of-sync and I've also noticed the PhatBox skipping the last few seconds of Ogg-encoded tracks from time to time.

I will happily live with these glitches, I was just hoping someone had found an ideal solution.  If no one has, I think I'll survive.

[Vince]

Vince?  Care to chime in?

I doubt a LAME padding tag would make mp3 playback gapless, the phatbox certainly wouldn't support that.  From what I understand, mp3 inherintly has gaps from filling out the last frame to a full frame length.  You'd have to fill it with the start of the next song for gapless playback.

Ogg is officially supported, I should update the site.  It's in the official firmware, and the official PMM builds support it.

Anything else I missed?

[sdc395]

I doubt a LAME padding tag would make mp3 playback gapless, the phatbox certainly wouldn't support that.  From what I understand, mp3 inherintly has gaps from filling out the last frame to a full frame length.  You'd have to fill it with the start of the next song for gapless playback.

As I understand it, the LAME tags (enc_delay and enc_padding) inform the decoder of how many samples to skip at the start and end of a file, allowing gapless playback where the player supports these tags.

Would support for these tags be a problem for the Phatbox hardware?