[axavier]

axavier: see if you have the new threaded file reader in your version - you can check by seeing if there's a filereader.cpp present in the xbmc/cores/paplayer folder.

if so, open up mp3codec.cpp in the same folder and look at the constructer - you will see it initializes the filereader class (m_filepap) with some default settings. depending on what version you have there'll either be 2 or 3 parameters.

in the case of 2 parameters, increase the second parameter to say 16 or so.

in the case of 3 parameters, increase the first parameter to 65536*16 or so.

see how it goes.

cheers,
jonathan

hello jonathan. i did the change in the file and did a compile today. i get less hiccups with the new parameter (1 or 2 a song ) instead of the 3 or four tha ti was getting before so there is definitely an improvement.

again the music cahce settings are set to default. do you want me to try anything else? also whats the relation sship between this parameter and the music cache. would it help if i increase it?

[jmarshall]

it depends on which version you had. the new version is the 3 parameter one.

increasing the first parameter in this case increases the total cache readahead we do.

increasing it will increase the amount of readahead in the cache. 1mb is the setting i gave you - try doubling it.

btw: have you tried other codecs to see if they behave the same way?

cheers,
jonathan

[axavier]

it depends on which version you had. the new version is the 3 parameter one.

increasing the first parameter in this case increases the total cache readahead we do.

increasing it will increase the amount of readahead in the cache. 1mb is the setting i gave you - try doubling it.

btw: have you tried other codecs to see if they behave the same way?

cheers,
jonathan

thanks jonathan i have the "3 parameter". let me do another build with the cache setting set to 65536*32 which will give me 2 megs of read ahead buffer.

i only have mp3's amigo. one other thing...how this buffers work with the system -> setting? my guess is they are not related and as a matter of fact paplayer uses those parameters not the ones in the system - > cache

[jmarshall]

exactly. paplayer does not use any of the cache settings at all - i only implemented the readahead class yesterday. eventually it'll be all hooked up once the bugs are ironed out and we have a solid implementation.

to be honest, with mp3 there's no reason why you'd need that much readahead - 2 mb is over a minute of data for most mp3's - it seems to me the issue lies elsewhere.

do you get anything output about the output buffers being low (in the logs?)

cheers,
jonathan

[axavier]

exactly. paplayer does not use any of the cache settings at all - i only implemented the readahead class yesterday. eventually it'll be all hooked up once the bugs are ironed out and we have a solid implementation.

to be honest, with mp3 there's no reason why you'd need that much readahead - 2 mb is over a minute of data for most mp3's - it seems to me the issue lies elsewhere.

do you get anything output about the output buffers being low (in the logs?)

cheers,
jonathan

nope but i have my logs set to to 1 level a ssoon as i finish doing the build i'll set them to 0 and test....

[jmarshall]

another thing: don't use pcm passthrough (audio settings), and see if disabling/enabling hq passthrough effects anything.

cheers,
jonathan

[axavier]

another thing: don't use pcm passthrough (audio settings), and see if disabling/enabling hq passthrough effects anything.

cheers,
jonathan

i have output stereo to all speakers using the high definition pack from microsoft. right now i'm using high quality resampling. with 65535*16 the hiccup occurs regardless of whether or nor i'm using high quality resampling. stereo is output to all speakers...

[axavier]

same result jonathan.....upgraded the buffer to 65535*32 and still the same result, altough not as frequent as 65535*4.

[sizzlechest]

has anyone tested gapless playback with the new buffer sizes?

[Aron Parsons]

still not completely gapless here. i still get hiccups as well.

[jmarshall]

please:

i need you to:

1. try as many different codecs as possible. supported are: mp3, mpc, ogg, flac, ape, shn.

2. get debug output (loglevel 0) for when it glitches.

3. check with pcm passthrough disabled.

4. if it glitches, also try tracks on the xbox harddisk and see if they glitch.

i can't reproduce, so can't do much without good reports.

cheers,
jonathan

[axavier]

has anyone tested gapless playback with the new buffer sizes?

there is a slight delay when i play them. foobar2000 reports the correct size of the mp3's i have tested several live albums with the same result (not to mention the hiccups).

[Aron Parsons]

should we disable "output to all speakers" as well?

[jmarshall]

otas shouldn't make too much of a difference - let me know if you find that it does ofcourse!

if it takes a while to load the file, then that's probably what's causing the gaps for you (output buffer runs dry). that'll be improved later.

i'm more worried about the glitches during normal file playback.

cheers,
jonathan

[axavier]

please:

i need you to:

1. try as many different codecs as possible. supported are: mp3, mpc, ogg, flac, ape, shn.

2. get debug output (loglevel 0) for when it glitches.

3. check with pcm passthrough disabled.

4. if it glitches, also try tracks on the xbox harddisk and see if they glitch.

i can't reproduce, so can't do much without good reports.

cheers,
jonathan

jonathan, i tried it with the logs set to maxximum logging and there is no report of problems here..... could it be a xbox model issue, i have a 1.6 version?

all my tests have been done by streaming the audio form my pc.... i'll do a couple of tests by putting the files in the xbox to see what happens.
what is the difference between the buffering used by paplayer and the one used by mplayer? it has been a while since the last time i developed something but it seems to me that one works better (mplayer's) than the other (paplayer), specially taking into consideration that video takes more resources to decode than audio.

jonathan, also is there an option to add more logging to paplayer (similar to the mplayer login option) in the xboxmediacenter.xml file?

anywya, just let us know what you need to to help you troubleshoot this issue ....