Viki

When I first started segmenting, I was told to make sure that the subs will pop up before I hear the voice. This of course is difficult. I tried to do this by watching closely the actors' mouth and cutting when I think they are about to speak. This did not work very well and I kept adjusting and refining. Needless to say, I took a long time (1.5 hrs) to segment a 7 min video. It was also exhausting.

Eventually I tried another technique which I find gives pretty decent segments without taking too much time. I suspect many experience segmenters are doing the same thing.

Simply cut immediately when you hear a voice. Based on my reaction time, I find that I am normally about 0.2-0.3 secs late. Immediately go back to the previous segment and adjust it upwards (Ctrl-Up Arrow) by 4-5 times. Then let the video continue until the voice stops and cut again. I try to cut about 0.4-0.5 secs after the voice stops. The main thing is that the segment must be long enough for the viewer to read the subs without having to pause the video.

When I do this, I rarely need to go back and forth to check my segments. After segmenting is complete, I will reload the whole video and check through again one final time.

Now I can segment a 9 min video in about 1/2 an hour.

Did you know that ANYONE can adjust a segment? I assumed Viikii allowed this so that subbers can adjust any out of sync segment. However, this leaves the segment vulnerable to vandals. So what can you do?

You can use the no-voice segment to help you. Fortunately, the no-voice segment is invulnerable. It can only be adjusted by authorized segmenters. So it acts as a hard wall which cannot be breached.

So when I encounter a long sequence of segments, I normally try to squeeze in some no-voice segments, even if it is 0.3 secs long. This provides some form of protection. I also like to believe it acts as a marker for the subbers.