Bug 1541 - AAC files "hiccup" during playback.
: AAC files "hiccup" during playback.
Status: RESOLVED FIXED
Product: Logitech Media Server
Classification: Unclassified
Component: Streaming From SlimServer
: 6.0.2
: Macintosh MacOS X 10
: P1 normal (vote)
: ---
Assigned To: Dan Sully
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2005-05-15 16:54 UTC by mc
Modified: 2011-03-16 04:18 UTC (History)
0 users

See Also:
Category: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description mc 2005-05-15 16:54:03 UTC
AAC files "hiccup" during playback; i.e., there are intermittent moments when
the sound cuts out. Also happens when searching the playlist.

No other processing is occuring in computer at the time, since the G4 Titanium
laptop is dedicated to slimserver.

This was not a problem in 6.0.1.
Comment 1 Dan Sully 2005-06-01 09:58:11 UTC
Is this with a Squeezebox1 or 2? Wired or wireless?

Are you bitrate limiting?
Comment 2 mc 2005-06-05 11:13:16 UTC
In response to 
------- Additional Comments From dan@slimdevices.com  2005-06-01 09:58 -------
Is this with a Squeezebox1 or 2? Wired or wireless?

Are you bitrate limiting?


This is a Squeezebox 1.
Wired (ethernet).
Not sure what you mean by bitrate limiting.

The skips in sound (2-3 skips in a row) appear to coincide with any change in
the menu (i.e., when song list updates between songs). Thus the preponderance of
skips at the start of songs.

These are Apple Lossless files that are set to "AIFF " decoding in "File Types."

NOT wireless.
Comment 3 Greg Cannon 2005-06-24 04:27:58 UTC
I also had sound corruptions after browser refreshes when streaming mp3s to xmms
or itunes clients. 

The stream is being trashed by the browser fetching status_head.html, as this
invokes the Slim::Player::HTTP::power method which calls
Slim::Web::HTTP::clearOutputBuffer() on the stream. 

Here's a snippet of d_http log output:
(note the browser is on a different box from the player, and I've added a few
extra d_http msg's)

2005-06-23 21:35:12.9207 Streamed 2896 to 192.168.1.3
2005-06-23 21:35:13.0189 sendstreaming response begun...
2005-06-23 21:35:13.0197 metadata bytes: 9145
2005-06-23 21:35:13.0203 sent incomplete chunk, requeuing 6623 bytes
2005-06-23 21:35:13.0208 Streamed 2896 to 192.168.1.3
2005-06-23 21:35:13.0517 reading request...
2005-06-23 21:35:13.0523 HTTP request: from 192.168.1.10
(HTTP::Daemon::ClientConn=GLOB(0x960a91c)) for GET HTTP/
1.1 /status_header.html?player=192.168.1.3&start=&refresh=1 
2005-06-23 21:35:13.0533 HTTP parameter player = 192.168.1.3
2005-06-23 21:35:13.0539 HTTP parameter start = 
2005-06-23 21:35:13.0544 HTTP parameter refresh = 1
2005-06-23 21:35:13.0556 processURL Clients: 192.168.1.3:32873
2005-06-23 21:35:13.0565 Generating response for (htm, text/html) status_header.html
2005-06-23 21:35:13.0571 generating from include.html
2005-06-23 21:35:13.0585 finished include.html
2005-06-23 21:35:13.0594 path status_header.html matches a page function:
CODE(0x8f63500)
2005-06-23 21:35:13.0607 Slim::Player::HTTP::power called
2005-06-23 21:35:13.0613 clearOutputBuffer on
HTTP::Daemon::ClientConn=GLOB(0x98801e8)
2005-06-23 21:35:13.0664 generating from status_header.html
2005-06-23 21:35:13.0707 End request: keepAlive: [23] - waiting for next request
on connection = keep-alive
2005-06-23 21:35:13.0718 More to send to 192.168.1.10
2005-06-23 21:35:13.0730 No more messages to send to 192.168.1.10
2005-06-23 21:35:13.0744 No segment to send to 192.168.1.10, waiting for next
request..
2005-06-23 21:35:13.1189 sendstreaming response begun...
2005-06-23 21:35:13.1200 (audio: 32768 bytes)
2005-06-23 21:35:13.1206 metadata bytes: 12041
2005-06-23 21:35:13.1210 splitting message for metadata at 20727
2005-06-23 21:35:13.1215 sent incomplete chunk, requeuing 17831 bytes
2005-06-23 21:35:13.1220 Streamed 2896 to 192.168.1.3

Editing Slim/Player/HTTP.pm power() sub to remove the call to clearOutputBuffer
seems to have cleared my problem. Not sure if thats the right or best thing to
do, but it seems to be mostly harmless. 

This may be affecting all clients that use the HTTP stream. I suspect AAC files
are less tolerant of this than mp3's, making the corruption more noticeable.

Hope thats useful.

regards,
Greg.
Comment 4 Dan Sully 2005-06-24 08:34:22 UTC
Thanks - this has been fixed as of subversion change 3532.