Annoying 'gap sound' when joining flv files

[KeyMs92]

Hi,

Yesterday I recorded part of a live stream, and the plan was to edit two clips out of it and subsequently join them. In order to do this, I used Ultra Video Splitter/Joiner for the splitting and joining, to avoid any loss of quality. Also, I used Avidemux to shift the audio by 180ms because of a desynchronization caused by the splitting process.

All good so far, except for one thing: the joining had left an audible gap between the two clips. Besides Ultra Video Joiner I tried a number of mp3 joiners (after extracting the mp3 track first) but they all failed. Then I noticed that when I tried to join the clips in Avidemux there was no gap. Pretty soon I realized why Avidemux was the only tool that worked: Avidemux detects a video framerate of 25,033 instead of the right framerate of 25,000 (also used by Ultra Video Joiner and indicated by MediaInfo). When you leave it at 25,033 the gap isn't present. However, it makes the video go too fast and thus cuts the end of the audio track. When you change it to 25,000 you hear the 'normal' gap sound. I also tried setting the framerate below 25,000 making the video go slower, but that causes the audio of the first clip to loop before the second clip starts (for obvious reasons).

All in all, it seems that for proper joining there's no way to avoid the gap sound. Of course, Audacity could be an option but that would mean encoding to mp3 for a second time.

I've two questions:

1. Is there a way to properly join the video files without the gap sound?
2. What would cause the gap sound to disappear when the framerate is set to 25,033? (If anyone knows.)

Thanks!

EDIT: To my utter surprise I noticed that these gap sounds are produced on single files as well, right at the beginning. In this case, they only happen when there's a delay specified, regardless of framerate. In the case of two files, they only happen when the framerate is not 25,033, regardless of delay. I double checked all scenarios to make sure. Also, I was wondering why the sounds are caused one single files because there's no transition. It turns out that when you specify a delay of e.g. 180ms the audio track already starts playing at 0ms, continues to +- 180ms and than starts again. The sound is caused by this restart transition. Lastly (to make it more confusing), it doesn't happen with all files...

EDIT2: To avoid having to use the delay, I tried joining 180ms silence with the track first. Guess what, it also causes the sound. I'm starting to get the impression it's just not possible without reencoding.

[Baldrick]

Try andys flv joiner. But I guess you wont be able to solve this that simple. Joining is a not as simple as it sound.

[KeyMs92]

Thanks... but nope, same old bleeping story

You're right about the joining, it's not quite as simple as one might think. I'm really starting to think that joining without a noticable sound really is impossible. I've tried dozens of joiners but they all produce the exact same result.

The only tool that really stood out for me was Virtualdub: It seems to offer the best options. Basically there are two methods. The first method is that you make a direct stream copy of the video stream and a full processing of the audio stream. The audio stream is then converted to PCM and any delay is then added to the stream (it adds actual silence). This opposed to when you make a direct stream copy of the audio stream (i.e. use an mp3 stream), the delay point is then defined in the container itself. Finally, the result is exported to avi (which must happen anyway when you choose PCM).

The first method requires the audio to be encoded again when you want to revert to mp3, but it works on all players. The second method seems ideal, but unfortunately it causes incompatibility problems with some players, causing the audio to be even more desynchronized.

Both methods are at least a lot better than what Avidemux has to offer (as mentioned in my first EDIT). As I'm uploading to Youtube, I chose the first method. Youtube has no support for a specified delay point and it seems to encode the audio to aac anyway. My guess is that the same method should be applied for joining (probably using Audacity in that case).