[Savonet-users] crossfade and liq_fade_out
Hi,
I need to insert commercials in a playlist.
To get tracks i use an external script to read next file from a playlist
and return metadata according to track type :
nofade for jingles/commercials and fade for songs. I send
liq_fade_in and liq_fade_out too.
My response from bash script (get.next.track.sh) is :
-
...
# for jingles
META=annotate:type=\nofade\,liq_fade_in=\0.1\,liq_fade_out=\0.1\,display_title=\TITLE\:/path/to/file.mp3
...
# for songs
META=annotate:type=\fade\,liq_fade_in=\0.1\,liq_fade_out=\3.\,display_title=\TITLE\:/path/to/file.mp3
...
# return response
echo ${META}
-
i use a crossfade function from examples:
def crossfade(~start_next,~fade_in,~fade_out,s)
s = fade.in(duration=fade_in,s)
s = fade.out(duration=fade_out,s)
fader = fun (a,b) - add(normalize=false,[b,a])
cross(fader,s)
end
radio = crossfade(start_next=1., fade_out=1., fade_in=1., radio)
==
Problem ONE
Now, at first impression, liq_fade_out is not correct. In almost all
cases, some 4-5 seconds overlap i hear.
But i tested intensively and is not from incorrect liq_fade_out handling.
In crossfade function, start_next parameter IS NOT USED !
And, in cross() function, duration is 5 seconds default.
If i add liq_start_next parameter to META (annotate), all is ok.
In this case, in examples, need to replace
cross(fader,s)
with something like
cross(fader,duration=start_next,s)
===
Problem TWO
if i use function
radio = crossfade(start_next=1., fade_in=0., fade_out=0., radio)
or
radio = crossfade(start_next=0., fade_in=0., fade_out=0., radio)
this is response from liquidsoap console log:
... Waiting for [EMAIL PROTECTED] to be ready...
... Waiting for [EMAIL PROTECTED] to be ready...
etc
this is WITH annotate liq_fade_in/out/start_next set like above, not to
zero.
10x,
Daniel
-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/
___
Savonet-users mailing list
Savonet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/savonet-users
Re: [Savonet-users] Multiple hierarchical sources limitations
Hi, Just a notice that Romain actually implemented kicking (as the stop command if I followed correctly) and per-mount passwords. See http://savonet.rastageeks.org/ticket/139 for details. On Mon, May 12, 2008 at 9:25 AM, Dimitris Nastos [EMAIL PROTECTED] wrote: Our DJ's use SAM Broadcaster to stream their music to Icecast. When there's nothing playing (i.e., temporary network congestion, no song queued up, no voice), SAM streams nothing, although it remains connected. Icecast does not disconnect, but liquidsoap gets the following error: 2008/05/12 00:27:23 [stream.mp3:2] Read error Unix.Unix_error(2, read, ) 2008/05/12 00:27:23 [/dj0:2] Feeding stopped: Mad.End_of_stream Which is kinda normal, but unless I have a way to define a timeout, liquidsoap will start jumping back and forth between live and playlist in case of temporary stream interruptions. Currently I am using a transition with a few seconds of blank to work around this. Regarding this issue, you can try playing with the buffer size parameters (cf. liquidsoap -h input.harbor) so that if the interruption is less than the buffer size, liq doesn't switch back to automated playlist. If that doesn't work, then I should try to understand why.. Anyway, buffering a lot of data might have its problems, in which case using a transition seems sensible. Cheers, -- David - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
Re: [Savonet-users] request.dynamic runs a script 3 times
Le Thursday 15 May 2008 23:52:39 Benoït Leudet, vous avez écrit : Hi, Hi ! It works well except that I was surprised to see that the script is called three times (two times if length=30.) and I would like that only one title is sent because I want to manage the playlist a more precise way. Is it possible or do I try another solution? Well, the script is called several times until the queue is fed, but the songs are played one by one... How does it cause trouble for you ? Romain -- How can a man Discover a land That already populated with Indians? - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
Re: [Savonet-users] crossfade and liq_fade_out
Hi, On Fri, May 16, 2008 at 9:46 AM, bede [EMAIL PROTECTED] wrote: To get tracks i use an external script to read next file from a playlist and return metadata according to track type : nofade for jingles/commercials and fade for songs. I send liq_fade_in and liq_fade_out too. First a small remark: the type=nofade/fade does not seem to be used in your script. Liquidsoap doesn't recognize anything like that: all it does is reacting to the liq_fade_* and liq_start_next metadata. def crossfade(~start_next,~fade_in,~fade_out,s) s = fade.in(duration=fade_in,s) s = fade.out(duration=fade_out,s) fader = fun (a,b) - add(normalize=false,[b,a]) cross(fader,s) end You made a good point about this function taken from our wiki: the start_next parameter is unused. In normal times I'd have advised you to fix it on our wiki, but now we're moving to another system for documentation, soon online. I'll fix it there. As you correctly said, the last line should be cross(duration=liq_start_next,fader,s). Now, at first impression, liq_fade_out is not correct. In almost all cases, some 4-5 seconds overlap i hear. The overlap is controlled by start_next, not fade_out. Cross_fade is only the duration of the fading, that is re-shaping of the enveloppe of an end of track. If i add liq_start_next parameter to META (annotate), all is ok. That makes sense. In the end, you understood the problem quite well, I'm just confirming everything you said :) Problem TWO if i use function radio = crossfade(start_next=1., fade_in=0., fade_out=0., radio) this is response from liquidsoap console log: ... Waiting for [EMAIL PROTECTED] to be ready... ... Waiting for [EMAIL PROTECTED] to be ready... etc Thanks a lot. You spotted a bug. The fade.out() operator fails to declare itself as ready when its duration is zero. I've fixed it on SVN. If you don't want to rebuild from SVN, you can safely use fade.out=0.0001 -- smaller than that may result in a rounding to 0. Have fun with liquidsoap! -- David - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
Re: [Savonet-users] Questions about delay and switch
David Baelde a écrit : I don't think you can get anything precise from icecast. Still imprecise, you can play with harbor's buffer: input.harbor(buffer=80.,...). A little less should be better actually, since other buffers come into play in the pipeline, before and after liquidsoap. I forgot to tell you that this is the solution we use and it works well with a buffer of 82. We just made our first shoutcasts with the delay and the results are very good. We need more tests to be sure. The only problem is we pray that the source will not be disconnected because if we reconnect immediately, we lost the sync. In this case, we have to wait 82 seconds to reconnect. OK, the shoutcast is cut but it's not dramatical. - This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/ ___ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
