Audio Sync After Stripping Tracks from Video

I’m using Audacity 2.0.6 with FFmpeg 2.2.2 on Windows 8.1.

I have a video file that’s a recording of a Skype video chat captured with Evaer. The audio is captured in stereo, so my mic is on one channel and the remote user is on the other.

I want to extract and separate the audio into separate files so that I can edit the audio, video, and video from another camera as 4 separate tracks.

So I import the mp4 file into Audacity and split the stereo to mono and export each track (with the other muted) into an MP3 file.

After this process I find that I can’t sync the audio and video because the extracted audio track is actually shorter than the original video (by maybe a couple seconds after an hour of video).

  1. Is this the correct process or is there a better way for me to accomplish this?
  2. Is there anything I can do to make sure the separate audio files match up better with the original video?

Thanks,
Mike

Avoid using lossy formats such as MP3 during the production process. Export as WAV so as to avoid quality loss at this stage in the process.

Have I got this right - you have 2 MP4 files, one from each end of the Skype conversation, and each MP4 has 2 channels of audio.So on one of the MP4s you have a direct recording of you and a remote (via Skype) recording of the other person, and on the other MP4 you have a direct recording of the other person and a remote (via Skype) recording of you. Is that right?

There is no guarantee that your MP4 recording will run at exactly the same speed as their MP4 recording, because the two cameras/computers are not synchronised. If I understand your setup correctly, your audio should synchronise OK with your video, and their audio should synchronise OK with their video, but to synchronise your audio / video with theirs, you may well need to tweak the speed a bit.

Effect > Change Speed. If you consistently record with the same equipment, you should only need to derive the error correction once. Once you know how far off everything is, it’s the same percentage no matter how long the show is.

One technique for a multi-microphone show is to have everybody wear tiny recorders. When all the sound files get to Edit Central for production, each one must be corrected for time duration. “If you were wearing the blue recorder, you’re going to be .004% fast…”

Never do production in MP3!! You can’t easily cut an MP3 into anything else. MP3 is a delivery to customer (not client) and final step.

Koz

Final Cut Pro has the ability to shrink and stretch videos very slightly. Your editor may need to do that to make everything line up. It’s not unusual for separate recorders to record slightly off sync. That’s why the grownups still have sync cables on a multi-camera show.

Koz

Close. It’s a single MP4 with 2 channels of audio; me on one channel and the person I’m talking to on the other.

I’m trying to split it out so I can match the audio of me with a completely separate video source, which matches the audio initially, but not after I export it.

I will try this first before I start trying to stretch or slow the audio or video.

OK. So what are the “4 separate tracks” that you refer to in your original post?

The video and two audio tracks from this recording are 3. The 4th is a video stream from a second camera that isn’t causing any trouble. (In testing the 4th track is irrelevant, so let’s continue without it. Sorry for confusing things initially.)

So, I tried exporting as a wav as suggested and now I’m more confused.

As an experiment, I dragged the original MP4 into audacity and export the tracks as mp3 and then as wav. I then brought the new files back in to compare wave forms and found the mp3 didn’t match, as expected, but the wav did, as hoped.

But then… I imported the MP4 and exported wav, which matched up in Audacity, into my video editor (Cyberlink PowerDirector) and they didn’t match up for length again. Since all the original files match up in PD, and the wav and mp4 match up in Audacity, I assume Audacity must be doing something slightly wacky with the mp4. Is there another conclusion I can draw?

Attached are the visual comparisons of the original mp4 and wav side by side in both PD (3.2 sec diff) and Audacity (no diff).

Any thoughts?


Or your video editor is - either way, MP4 is a pretty wacky format.

Rather than importing the MP4 into Audacity, try exporting the audio from your video editor in WAV format, and import that WAV file into Audacity.

Exporting the audio from the video editor and then using Audacity to split left and right seems to work fine.

I have a process now. Thanks, Steve!

I assume Audacity must be doing something slightly wacky with the mp4. Is there another conclusion I can draw?

I actually think it’s something with the MP4 file. I’ve sometimes had trouble with the audio & video going out-of-sync with editing MP4 and (other highly-compressed formats) with my video editor.

Sometimes, I’ve been able to “fix it” by converting the file to MPEG-2 (usually with uncompressed LPCM audio) before editing. Unfortunately, this means the video is going through an additional lossy compression step.

(MP3 compression will add a few milliseconds of silence to the beginning of the file.)