Re: [Debconf-video] MiniDebConf Prishtina video
On 12/10/17 08:17, Wouter Verhelst wrote:
> On Wed, Oct 11, 2017 at 09:32:46PM +0200, Wouter Verhelst wrote:
>> On Wed, Oct 11, 2017 at 09:30:26PM +0200, Wouter Verhelst wrote:
>>> On Wed, Oct 11, 2017 at 09:02:02PM +0200, Wouter Verhelst wrote:
>>>>>>> Can you suggest how volunteers can retrieve and play with these files
>>>>>>> now? I'd like to notify them on the FLOSSK list (Kosovo) and Open Labs
>>>>>>> forum (Albania)
>>>>>> I'm working on the necessary changes for sreview, and then people will
>>>>>> be able to manage them through there. For that though, I'd need to know
>>>>>> what the start time (preferably at as small a resolution as you can
>>>>>> manage) of the different WAV files is. Currently that's kindof unclear.
>>>>>>
>>>>>> If you want to absolutely not do that, then hrm, we'll have to think of
>>>>>> something else.
>>>>>>
>>>>>
>>>>> I think the WAV filenames are UNIX timestamps
>>>>
>>>> Ah, yes, now that you mention it :-)
>>>>
>>>> (they're a bit big, but I presume that's because they have milliseconds
>>>> in them -- awesome)
>>>
>>> Now that I've imported all the raw assets into the database, I notice
>>> that my guess that the mtimes on the files would be the end time of
>>> their recording seems to be incorrect, too:
>>
>> Actually, nvm, I'm an idiot, you said it's in the metadata, and that's true.
>> Ignore me please :-)
>
> It would have been nice if the timestamps would have been at least remotely
> correct :-(
>
> sreview=> select * from raw_files where room >= 6 order by starttime, room;
> id | filename | room | starttime | endtime
> ------+-----------------------------------------------------------+------+----------------------------+----------------------------
> 2040 | /home/pocock/2017-10-07-prishtina/audio/1507366756539.wav | 7 | 2017-10-07 10:59:16.539+00 | 2017-10-07 13:14:12.939+00
> 2065 | /home/sreview/prishtina/DSC_6640.MOV | 6 | 2017-10-07 11:13:07+00 | 2017-10-07 11:18:30.56+00
>
> Yet if I download those files to my laptop and play them, it turns out
> they're within about a minute from one another. SReview does allow you
> to enter A/V sync correction values, but it does so by throwing away a
> bit of content at the beginning of the "pre" recording. This isn't so
> much of a problem if you're only correcting a few seconds (since the pre
> recording is just for context anyway), but in this case we might
> actually be throwing away entire talks. Obviously that's not the idea.
>
> Additionally, neither of the above two lines is correct according to the
> schedule; this is the opening speech, which was scheduled to start at
> 10:30. You did say that you didn't follow the schedule completely, which
> is fine, but this feels odd, and I'd like to make sure that I'm not
> entering the wrong information so that reviewers will really have to do
> a lot of work...
>
> Can you give me an offset for both? That way, I can update the times in
> the database and then hopefully things will look much better.
>
Are you looking for the offset between the camera and the audio, or the
offset from the schedule?
The event started later than scheduled and finished earlier. Some talks
were a lot shorter than scheduled, some of them start later than they
should and some start earlier.
Regards,
Daniel
Reply to: