Re: [Mlt-devel] melt source.txt color:blue

[Brian M. <pez...@ya...>]

>>  Also, if you are testing lipsync, have you looked at the blipflash producer 

> and consumer? The producer generates a synchrinized blip and flash pattern. And 
> the consumer can calculate the lip sync offset given that pattern as an input:
>> 
>>  melt -silent -profile dv_ntsc blipflash out=200 -consumer libdv:temp.dv 
> terminate_on_pause=1
>>  melt -silent -profile dv_ntsc libdv:temp.dv -consumer blipflash 
> terminate_on_pause=1
> 
> Oh yeah, I heard about this, never got around to trying it till just
> now.  very cool - thanks.
> 
> as long as we are doing so good on the quickies here...
> 
> What i am currently testing for is gaps in audio so I need more than
> the blip.  can I get mix the
> "producer noise" with it?

It would look something like this:
melt noise -track blipflash -transition mix a_track=0 b_track=1

The blipflash consumer should be able to detect the blips as long as the gain on the noise isn't too high.

> It is OK if that confuses the blipflash test, so far sync is on the
> order of 10 seconds after well over 12 hours, (currently the steam
> gets reset every 4 hours, and no one has noticed the sync problem)
> but the audio drops are for a few seconds when I transition between
> sources - very noticeable and disruptive.   (this is a problem in
> flumotion/gstreamer, not mlt)

Sine wave test tones are very useful for detecting audio frame loss because when an audio frame is lost, it creates a discontinuity in the tone that is easy for the human ear to detect. The ladspa producer can produce arbitrary sine waves if you need it. Good luck finding your problem.

~Brian