Details
Improvement
Closed
Major
Description
Blank detection is a must as we don't uses pre-cut mp3 files to upload to airtime ( i have a feeling more people don't ). Most mp3 files contain blanks end early and or end to silent. When that happens airtime should skip to the next song.
We been tweaking the liquidsoap on our streaming server to sort out blanks etc. Where currently satisfied with the script where running there. It uses s = skip_blank(length=2.,s) which works like a charm and skips to the next song when the song has to mutch blank.
Now we tried to incorperate that into airtime, becuase we noticed to mutch blanks there as well. So we added the same to the liquidsoap script, which goes wel for a while. After a few skips the whole thing goes haywire though.
Songs displayed in Airtime are not the ones currently playing, and after some skips airtime stops playing totally for some time. The biggest question, is ther ea way to implement the blank detection or am i doing that at the wrong point ? Is there a way to make it more likable for airtime ?
We been tweaking the liquidsoap on our streaming server to sort out blanks etc. Where currently satisfied with the script where running there. It uses s = skip_blank(length=2.,s) which works like a charm and skips to the next song when the song has to mutch blank.
Now we tried to incorperate that into airtime, becuase we noticed to mutch blanks there as well. So we added the same to the liquidsoap script, which goes wel for a while. After a few skips the whole thing goes haywire though.
Songs displayed in Airtime are not the ones currently playing, and after some skips airtime stops playing totally for some time. The biggest question, is ther ea way to implement the blank detection or am i doing that at the wrong point ? Is there a way to make it more likable for airtime ?
Activity
Hide
Paul
added a comment -
Permalink
User should be able to turn this on and off in the preferences and set what the blank detection threshold should be.
Show
Paul
added a comment - User should be able to turn this on and off in the preferences and set what the blank detection threshold should be.
Stripping silence automatically within files could be a problem if that silence is meant to be there - for example, in classical music where people don't applaud between movements:
http://www.nytimes.com/2008/01/08/arts/music/08audi.html?_r=3&oref=slogin&oref=slogin
It might be better to limit silence stripping to the beginning and end of the file only.
We could also offer a warning on upload such as "2.3 seconds of leading silence stripped from <file>" or "1.4 seconds of trailing silence stripped from <file>".
http://www.nytimes.com/2008/01/08/arts/music/08audi.html?_r=3&oref=slogin&oref=slogin
It might be better to limit silence stripping to the beginning and end of the file only.
We could also offer a warning on upload such as "2.3 seconds of leading silence stripped from <file>" or "1.4 seconds of trailing silence stripped from <file>".
Show
Daniel James
added a comment - Stripping silence automatically within files could be a problem if that silence is meant to be there - for example, in classical music where people don't applaud between movements:
http://www.nytimes.com/2008/01/08/arts/music/08audi.html?_r=3&oref=slogin&oref=slogin
It might be better to limit silence stripping to the beginning and end of the file only.
We could also offer a warning on upload such as "2.3 seconds of leading silence stripped from <file>" or "1.4 seconds of trailing silence stripped from <file>".
See this forum post:
http://forum.sourcefabric.org/discussion/14165/feature-request-silence-detectionremoval-in-recorded-audio
Here's how to use SOX to strip out silence:
http://digitalcardboard.com/blog/2009/08/25/the-sox-of-silence/
http://forum.sourcefabric.org/discussion/14165/feature-request-silence-detectionremoval-in-recorded-audio
Here's how to use SOX to strip out silence:
http://digitalcardboard.com/blog/2009/08/25/the-sox-of-silence/
Rather than modify the audio, we could use silence detection to find the nearest zero crossing before the start of the audio, and the nearest after the audio ends, and insert cue in/cue out points into the database for that file. We might need a threshold above absolute silence to cope with low-level background noise.
Show
Daniel James
added a comment - Rather than modify the audio, we could use silence detection to find the nearest zero crossing before the start of the audio, and the nearest after the audio ends, and insert cue in/cue out points into the database for that file. We might need a threshold above absolute silence to cope with low-level background noise.
Show
Daniel James
added a comment - Robin has written a silence detector:
http://gareus.org/gitweb/?p=silan.git
testing instruction:
1. Download and compile silan. (get it from the link that Daniel provided)
2. Make a symlink under /usr/bin/ ( run 'ln -s <path to silan> silan' inside /usr/bin/)
3. test it.
1. Download and compile silan. (get it from the link that Daniel provided)
2. Make a symlink under /usr/bin/ ( run 'ln -s <path to silan> silan' inside /usr/bin/)
3. test it.
Show
James Moon
added a comment - testing instruction:
1. Download and compile silan. (get it from the link that Daniel provided)
2. Make a symlink under /usr/bin/ ( run 'ln -s <path to silan> silan' inside /usr/bin/)
3. test it.