silence between songs - Airtime Support Discussions on Sourcefabric Forum

Bart · Post edited by Bart at 2013-02-28 04:38:49

Hello All,

After leaving airtime a while ago, because of important missing features, I now installed Airtime 2.3.0

I now have a question, of course :)

If I make a playlist, the songs are being played, which is good of course.

But after each song there is always some small silence, for starting the new song, which doesn't sound professional.

Is there a way to remove that "silence", so that there is a smoother transition to the new song.

Bart

Albert FR

2.3.1 ???

Bart

;)

2.3.0 I mean ..The 1 key is very close to my 0 key on my keyboard ;)

Albert FR

lol

these silence come on airtime playlist or playlist generated into ls_script.liq ?

Bart

airtime playlist. I did not modify ls_script.liq

Martin Konecny

2.3.0 comes with a feature that automatically set cue points on your system based on whether leading/trailing silence. You can verify this by clicking the >[ ]< icon in your playlists and seeing if the cue_out point is less than the length of the track.

Did you install from the Debian package?

Bart

Hi Martin,

On debian, the install failed ;)

I'm using the install on Ubuntu now, using the standard ubuntu/debian package. (which is a fresh clean system)

I can see the cue points, it's almost always the complete duration or even higher.

For example:

Song 1, duration: 4:55.7

cue out: 4:55.76

Song 2, duration: 6:28.4

cue out: 6:28.49

Song 3, duration: 9:25.2

cue out: 9:25.45

Martin Konecny

@Bart,

Can you open a terminal and try running "silan --version"

Bart

Hi Martin,

#>silan --version

silan version 0.3.0

Copyright (C) GPL 2012 Robin Gareus <robin@gareus.org>

Martin Konecny

What's the output of the following two commands:

sudo -u postgres psql -c "select count(*) from cc_files where silan_check = 't'" airtime

sudo -u postgres psql -c "select count(*) from cc_files where silan_check = 'f'" airtime

?

Bart

1)

count

-------

42

2)

count

-------

0

Helmut Müller

Hi Bart,
do you mean crossfading? Then you could change this in your ls_script.liq. But then everything is crossfading.
I would love to have the possibility to choose on every file "Crossfading: Yes, No". So i could make every Song crossfading. News, Jingles, Ads no crossfading. Would be a very nice feature.

Martin Konecny

@Helmut,

We are planning on crossfading support very soon. We already have some
ideas of our own, but would love to hear feedback from the community on the
desired workflow.
On Mar 1, 2013 1:03 PM, "Helmut Müller" <<br />airtime-support@lists.sourcefabric.org> wrote:

> Hi Bart,
> do you mean crossfading? Then you could change this in your ls_script.liq.
> But then everything is crossfading.
> I would love to have the possibility to choose on every file "Crossfading:
> Yes, No". So i could make every Song crossfading. News, Jingles, Ads no
> crossfading. Would be a very nice feature.
>
>

Bart

so...is it now "normal" or not...

Helmut Müller

It is normal. You can try to change your ls_script.liq

queue = crossfade_airtime(queue)

to

queue = smart_crossfade(queue)

But timing (schedule) will not be exactly anymore.

@Martin
Great to hear. maybe we could organize the files in "Type" e.g. Song, News, Advertisment, Jingles etc.

Mario Costa · Post edited by Mario Costa at 2013-03-03 08:49:41

Hi,

I've noticed that, if I insert a playlist that all use blocks, the cue out value is exactly as the file.

But, If I insert manually this file in a Playlist, the cueout value is different

SELECT length FROM cc_files WHERE id=1219

length
00:03:40.461979

SELECT DISTINCT cue_out FROM cc_schedule WHERE file_id ='1219';

cue_out
00:03:40.461979
00:03:40.4

If you check in the cc_playlistcontents table, you'll see that cueout is correct.

id	playlist_id	file_id	block_id	stream_id	type	position	cliplength	cuein	cueout	fadein	fadeout
455	12	1219	NULL	NULL	0	27	00:03:40.461979	00:00:00	00:03:40.461979	00:00:00.1	00:00:00.1

Please note that, the Cue_out value of "00:03:40.4" is only set on schedule if I insert the file manually in a Playlist.

When I add the playlist to the show, the cueout will be: 00:03:40.4

If that file (1219) is added to a show by a Block, then the cue_out is: 00:03:40.461979

I believe this is making some small gaps of silence because, normally I use Blocks and, no problem at all.

When I manually build a playlist, I notice small gaps of silence...

Can this be fixed?

The cue_out value when manually inserted?

Many Thanks

Martin Konecny

Bart said:
so...is it now "normal" or not...

Your database values look fine. How long exactly is the gap between tracks since they should be almost eliminated. As I said in the other post though, 2.4.0 will allow overlapping tracks (crossfading) so that even split seconds of silence are eliminated.

Roland Schwarz

Martin Konecny said:
2.3.0 comes with a feature that automatically set cue points on your system based on whether leading/trailing silence.

This is a handy feature most of the time! But how shall I proceed if I do not want this for some particular files? E.g. I need som silence either at the end of the last or the beginning of the next file to account for a small momnet between a time signal and speaker announcing.

Roland

Bart

sorry, coming back to this one, using 2.4.0, I'm having those results:

--
sudo -u postgres psql -c "select count(*) from cc_files where silan_check = 'f'" airtime
count
-------
772

(1 row)
sudo -u postgres psql -c "select count(*) from cc_files where silan_check = 't'" airtime
count
-------
2739
(1 row)
--

if I see this big difference, does that mean that silan still need to do his things?
If so, what is the best way to 'monitor' (logfile?) the progress of silan?

Cliff Wang

Silan log can be monitored by tail -F /var/log/airtime/pypo/pypo.log, it should happen every 5 mins

Bart

yeah, I found the reason already (not yet the root cause), and I've opened a ticket for this:

http://dev.sourcefabric.org/browse/CC-5270

The system loops on invalidcontenttype:
2013-07-10 15:20:25,083 ERROR - [silananalyzer.py : main() : line 72] -

2013-07-10 15:20:25,083 ERROR - [silananalyzer.py : main() : line 73] - Traceback (most recent call last):

  File "/usr/lib/airtime/pypo/bin/media/update/silananalyzer.py", line 70, in main

    self.api_client.update_cue_values_by_silan(processed_data)

  File "/usr/lib/airtime/api_clients/api_client.py", line 491, in update_cue_values_by_silan

    return self.services.update_cue_values_by_silan(_post_data={'data': json.dumps(pairs)})

  File "/usr/lib/airtime/api_clients/api_client.py", line 150, in __call__

    raise InvalidContentType()

InvalidContentType

2013-07-10 15:20:25,083 INFO - [silananalyzer.py : main() : line 75] - Processed: 100 songs

Cliff Wang

Have you checked cc_files for the cue in/out values? Maybe something is wrong there.

Bart

Hi Cliff,
Yes I did, but difficult to see if there is 'something' wrong over there, as that
table has +/-4500 records. sorting on cue in/out values didn't highlight anything.
somehow I need to find out which one is blocking the system , but I didn't find a way yet.

Cliff Wang

Understood.

Please backup the file /usr/lib/airtime/pypo/bin/media/update/silananalyzer.py and use the attached file to replace it (rename it from .log to .py).

Then restart airtime-playout service and post /var/log/airtime/pypo/pypo.log around that exception when it happens again.

Bart

Hi Cliff,

Tnx for your support already.
See attached the logfile.

Bart

Cliff Wang

Interesting item: [(2372, {'cueout': '287.422426', 'length': '173311.111837', 'cuein': '0.724218'})

Could you take a look at cc_files:

select * from cc_files where id=2372

Is it really long as 173311 secs?

Bart

Hi Cliff,

2372audio/mp3audioclip4Frans/08. Patrick Bruel - Casser La Voix.mp3empty0NULL2013-07-11 09:14:052013-07-08 02:15:47NULLNULLPatrick Bruel / Casser La VoixVarious16000044100NULL00:04:50.821224Joe FM De Franse Top 100FranstaligNULL20135NULLNULLNULLNULLNULLNULLNULLNULLFransNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLNULLTRUENULLNULLNULLNULLNULL-7.190000100:00:0000:04:50.821FALSEFALSEFALSEFALSE

id	name	mime	ftype	directory	filepath	state	currentlyaccessing	editedby	mtime	utime	lptime	md5	track_title	artist_name	bit_rate	sample_rate	format	length	album_title	genre	comments	year	track_number	channels	url	bpm	rating	encoded_by	disc_number	mood	label	composer	encoder	checksum	lyrics	orchestra	conductor	lyricist	original_lyricist	radio_station_name	info_url	artist_url	audio_source_url	radio_station_url	buy_this_url	isrc_number	catalog_number	original_artist	copyright	report_datetime	report_location	report_organization	subject	contributor	language	file_exists	soundcloud_id	soundcloud_error_code	soundcloud_error_msg	soundcloud_link_to_file	soundcloud_upload_time	replay_gain	owner_id	cuein	cueout	silan_check	hidden	is_scheduled	is_playlist

The file 5 minutes+

Bart

In CSV is easier readable I think :)

Cliff Wang

Can you share that file to me?

BTW, I'm wondering if you simply mark that file as "silan_check=t" will fixed your problem.

Just do

update cc_files set silan_check='t' where id=2372;

Bart

Can you grab it from here please: http://37.252.124.212/file.mp3

Cliff Wang

Thank you, I get the file.

It looks like this file cause the problem. Did you try the solution I mentioned?

Bart

the solution indeed worked, as workaround.

Cliff Wang

Ok.

Please also replace the silananalyzer.py cause it's debug version. Ticket for this problem is created.

http://dev.sourcefabric.org/browse/CC-5271

Howdy, Stranger!

Categories

Poll

Top Posters