Gapless playback not working in latest Ipod video builds?

[squidkidd]

Hey All,

I just recently bought a new-to-me 5.5g 80gb ipod off ebay after losing my old one and installed rockbox 3.9.  I'm having an issue I never experienced on my old ipod - if the disk is not actively spinning at the time playback reaches the end of a track there will be a gap between tracks with a slight audible 'hiccup' as the disk spins up.  Essentially, the disk must be up for playback to move on to the next track.  This happens with equalizer both on and off.  I've also played around with disk spindown settings to no avail.  I've seen this behavior now on 3.9 and also the current build r30121.

I went back and installed 3.8 and the problem is no more.  What gives in the new version? 

Regards,
Brian

[torne]

Sounds like your anti-skip buffer is set too small. Try increasing the setting.

The change in behaviour is probably because DMA was disabled before the 3.9 release.

[Buschel]

Torne, if you are talking of ATA DMA: It was disabled in 3.8 (or 3.8.1) as well.

[torne]

It was only disabled in 3.8.1, not in 3.8.

[Buschel]

@squidkidd: Torne points to an interesting option: To check if the the DMA change is the root cause you might just test 3.8.1 vs. 3.8.

[squidkidd]

Thanks for the replies,

It may be a few days before I can test this.   I should note though that on 3.9 I also changed the anti-skip buffer up to 5 minutes and still noticed the behavior.  What is DMA though and why was it disabled?

Regards,
Brian

[saratoga]

It's a disk mode that seemed to cause problems on some fraction of 5G ipods.

[dreamlayers]

I don't don't this is related to disabling of DMA. I'm not even sure it's related to buffering. The watermark calculation seems fine, and anyways, the disk shouldn't have to be accessed when switching tracks. It would only need to be accessed if by coincidence the buffer got to the watermark at that time.

[xenon]

Maybe 3.9 is not working gapless playback for AAC encoded by nero 1.3.3.0 and FAAC.

[Lear]

The gapless handling in the AAC decoder was recently changed, to make it work properly for files encoded by iTunes. It is possible it broke things for Nero files. I'll take a look at it.

[torne]

If the gap disappears if the disk is already spinning at the time then it can't be the fault of the gapless code; it must be buffering/diskio to blame..

[saratoga]

The gapless handling in the AAC decoder was recently changed, to make it work properly for files encoded by iTunes. It is possible it broke things for Nero files. I'll take a look at it.

I thought we didn't support nero gapless?

[Lear]

If the gap disappears if the disk is already spinning at the time then it can't be the fault of the gapless code; it must be buffering/diskio to blame..

But if a huge anti-skip buffer doesn't help (unless that's what is broken)? And the AAC decoder was changed in 3.9, so it is worth a quick look at least. I only tested with some iTunes/QuickTime files when I made the change...

I thought we didn't support nero gapless?

We do. Search for MP4_chpl in metadata/mp4.c.

[Lear]

And the AAC decoder was changed in 3.9, so it is worth a quick look at least. I only tested with some iTunes/QuickTime files when I made the change...

Sounds fine to me using Nero 1.5.4.0, when testing a few tracks.

[xenon]

Sounds fine to me using Nero 1.5.4.0, when testing a few tracks.

Nero 1.5.4.0 write iTunes compatible gapless data. (write iTunSMPB to metadata.)
>http://www.nero.com/enu/release-notes-nerodigital-nero-aac-codec.html

Please test using FAAC ?