Skipping Ogg Files

[Denzyl]

Hi All

I have been using Rockbox on my X5L 30Gb since 20-Aug-06

At first I used the Builds directly off the Rockbox site!

But since December 2006 I started to use the Maxwen builds due to the WPS patches!

I don't download every version that comes out though!

Anyway as of the 7th March 2007 build Ogg files started skipping usually about 20 to 30 seconds after start, the skip is usually about a half beat to single beat in length. They then continue to skip every 20 to 30 seconds throughtout the playing of that file!

I also tried the 11th March 2007 Maxwen, still skipping!

So I swapped back to a pure 'Rockbox' mainsite version latest build 19th March 2007 (r12841 - 070319) skipping seems more pronounced & noticable (But this might be due to me now knowing what to listen for!)

Now for details! All Ogg files are Winamp 5.2 to 5.33 rips direct from CD's I own!

Details from a skipping Ogg file

Length : 5:43
Average bitrate : 141 kbps
File size : 6,076,031 bytes
Nominal bitrate : 160 kbps
Channels : 2
Sampling rate: 44100 Hz
Serial number: 26135
Version : 0
Vendor :
Xiph.Org libVorbis I 20040629

Playback of the above file on Winamp and the pre March Rockbox version was flawless!

I reset all Rockbox config files before I reboot to new version!

Rockbox Bootloader is version 2.0

[Lear]

Strange. It sounds like Rockbox doesn't decode the audio data quickly enough, but for such an Ogg, that shouldn't happen. If you go to System > Debug > View audio thread, you can see if that is the problem.

The interesting values are mainly cpu freq and boost ratio. If everything is normal, the cpu freq should alternate between two values (being at the lower value most of the time) and the boost ratio should be around 30 for Ogg files with that bitrate (assuming the X5 CPU is about as fast as the one in the H140).

[Denzyl]

Thanks for the help Lear.

Here is the debug data from the Ogg in my first post!

CPU Frequency is 45Mhz to 124Mhz  as you predicted 124Mhz is very fleeting (it took 3 attempts to actually be able to read it) and it is mainly 45Mhz

CPU Boost started at 0% jumped to 52% then swings between those figures, heading down to a final 16% after about 10 seconds and is rock solid after that at 16%

pcm: is about 40,000 to 50,000 out of a Max of 529200

While the codec is running 6,000,000 out of 12,765,152 But occasionally jumps to Max then falls back to around 12,000,000 and stays there for a while slowly falling back to 6,000,000!

Pcmbufdesc runs up to 15/21 then fluctuates between 12/21 & 15/21

Hope this helps

[Llorean]

And was your audio skipping during this time period?

The PCM buffer should be at 0 when the audio skips under normal behaviour (well, under "the most common cause for skipping" style behaviour).

[Lear]

The PCM buffer should be close to full, so this sounds like the problem, but I don't really know what's causing it. It almost like it is stuck in low-latency mode... Did you have any sound-processing options on, like the equalizer or crossfade? If so, have you tried without them enabled?

[bascule]

I reset all Rockbox config files before I reboot to new version!

Do you mean you actually did a Reset Settings from the menu? If you have accidentally got stuck in low-latency mode, then a proper reset should work to fix it.

[Denzyl]

Hi All Thanks for the input

Llorean

I was listening to the track at the time and it was skipping!

Lear

Watching the Debug the skips definitely coincides with the drop in the pcm line!

I normally don't use any of those as I prefer seamless joins! (a lot of concept albums & comedies where gaps between tracks is a pain) It's the reason I went for Rockbox in the first place the original firmware always gave a large gap.

Bascule

I normally delete all the old config files on updating to a new version. Including all those generated by the previous system. Then copy in the new ones from the update, then let Rockbox regenerate the others on reboot.

This time I also tried the 'Reset Settings' from the 'Manage Settings' menu!

It got worse in that the skips now last longer, almost 2 beats. I am also now getting repeatable double skips of a 1 beat skip followed quickly by another one?

So I've switched off and tried a hard reset with a paperclip!

No improvement. Overall now on the same track the pcm is now definitely pumping (On skips it drops dramatically then goes to max, especially the double skips) and the 124Mhz is lasting long enough to be able to read it easily.

I'll check the Rockbox manual to see what I need to do a proper reset!

[nls]

Resetting settings in the menu or deleteing the config.cfg file is a proper reset. What seems strange about this is that usually the cause of similar skips is that the processor can't decode audio fast enough, which would mean that the boost ratio would be 100%, now this is not the case.
Does it happend on all tracks?
Do the tracks play correctly on your computer?
Have you tried older dailies to try and track down when this started?
After resetting settings could you try without changing any settings?

[Denzyl]

Hi nls

Ah! thanks for that on reset!

Okay I have now tried a few more tracks most with an obvious beat, they skip! Those files that are spoken I cannot tell!

All copies of the same files on my PC play okay on my PC! With Winamp & Media Player Classic!

I found a 16-August-2006 version of Rockbox lurking on my PC and loaded it on my X5L

The skipping disappears from most tracks, totally! But now I am attuned to it there is a residual skip on really hard beat tracks?

One other thing is there is a size difference between the Vorbis Codec files!

Okay I have found an interesting quirk!

Playing a file on the X5L over the USB link using Winamp or MPC on the PC it skips!
Copy the X5L file back to PC and then play it with Winamp or MPC, it doesn't skip!

I will check my X5L disc integrity!

[danhibiki]

The problem you're having sounds pretty much what I am having with my iPod Nano except mine doesn't start so quickly and I didn't notice if those problems match the pcm buffer decreasing.
I described it here: http://forums.rockbox.org/index.php?topic=9186.0
Can you test with some MPC or MP3 and see if you have the same issue reported there?

[Denzyl]

Hi All

danhibiki

I have very few MP3 files. But tests on the ones I do have show no skipping, using any of the 3 Rockbox systems I have 19-March-07, 07-March-07 & 16-August-06. I tested this before I started to defragment my X5L HD.

As for my discovery of the skipping when playing over the USB! I have checked the X5L Hard Disc.

Partion Magic showed no errors!
Microsoft Scandisk showed no errors

PerfectDisk showed 103 fragmented files mainly Rockbox system files and wps bitmaps. But also included the Vorbis codec.

So I ran the defrag, the result was interesting, all skipping stopped!

I have now reloaded the Rockbox 19-March-07 system! Then defragmented it!

The results are no skipping! Including playing the file over the USB link by Winamp on the PC!

The Audio Thread debug run on the Ogg file that I tested the first time now gives these results!

The codec is running between 10,000,000 & 12,000,000 out of 12,765,152 falling to a low of 9,000,000 at end of track play

pcm is bouncing between 390,000 & max out of 529,200

pcmbufdesc is pretty steady at 14/21 to 15/21 and switches a lot less than it did before!

CPU frequency is 45Mhz with flashes of 124Mhz

Boost ratio is a constant 16% with no fluctuation like previously at start up, it now goes straight to 16% and stays there.

The only other thing to note was that none of the ogg files that I used for testing were fragmented!

So it looks as if the fragmented system files were the problem!

I am at a bit of a loss over why the tracks skipped when played over the USB link by Winamp on the PC if it was fragmented system files causing the problem?

If I get time I will see if I can deliberately fragment the system files again to see if I can get the tracks to start skipping again!

[Denzyl]

Well trying to deliberately fragment files it was harder than I thought, so I gave up!

But as it turns out I think I found the underlying reason for the skips from another accidental discovery!

As part of my attempts to see if it was fragmentation and the placement of system files I used Norton Speed Disc 2005 to move all .Rockbox files to the start of the disc & make sure all music came after it (But not at the end of disc)

The result was unusable!

Rockbox booted slightly faster, but all the music tracks were now unplayable?

The track I have used in all my tests, now had 2 to 3 second leaps, stuttering and silences!

Fearing it had corrupted the file I checked the X5L disc, all secure & uncorrupted. So I dragged the file back to the PC and it played okay?

While looking deeper I found that the directory structure on the X5L disc was very fragmented, I suddenly realised that the problem was due to the disc hunting for files & fragments

So my original problem was minor skips!

Files that had been updated repeatedly, had fragmented and their directory entry was no longer near the file as well.
This meant the disc had to find directory then the file parts = skipping!

Speed Disc meant now that no file was near it's directory entry and in some cases they were at either end of the 23Gb of files I have on the X5L, as this now included the all the music files and folders as well! = Major problems as the disc tried to keep up with X5L's demands!

As a result I re-formatted the X5L drive using the command-line FORMAT command /x & /u. Reloaded the latest Rockbox then added the music.

The result is a rock steady with perfect playback. So I will stop fiddling (for now )

Thanks again for all those who answered my original queries, I hope that this insight might be of use, if it is correct of course