Sampling from a specific audio channel of a movie

While working on my last track I ended up working out some methods for how to rip samples from a movie, and access each separate channel rather then mixing down into a stereo signal prior to "recording". This method also skips the need to record anything, we can access the original audio files directly and avoid any signal degradation.

The reason this can be beneficial is due to how audio is generally stored on a movie. A “5.1” channel movie means you have 6 channels, there is your standard left and right speakers, your left and right surround (back) speakers, your sub/bass, and the key ingredient here, the center channel.

The center channel is generally used for voices, so by being able to sample from specific channels, we can now isolate to just the vocals, or remove the vocals from the mix depending on what we are trying to accomplish.

I'm not looking to create an in depth guide here. Just wanted to share a technique I worked out and the basics of what's involved. This is specific to movies in MKV format, which is a common format found these days, but the same concept could generally be applied to other movie types if you get the appropriate tools to demux the file.

Tools Used:
Audacity 1.3.13 (Beta) - Make sure to get the beta, not the stable
Ffmpeg For Audacity - To open DTS/AC3 files.
MKVtoolnix - To split audio from MKV files
MKVExtractGUI2 - GUI for demuxing in MKVtoolnix

1. Install Audacity
2. In Audacity you need to install FFmpeg. Goto Preferences->Libraries, you get the download link from there and then you just set the directory once it’s installed.
3. Install MKV tools
4. Run MKVExtractGUI2 and pick your MKV file
5. The idea behind this program is that will take your MKV file and separate it into all it’s parts. Once you pick the file, all the parts (Video, Audio, Subtitles, etc) should appear in the window with checkboxes. We only need the audio so we check just this. If the movie has multiple languages you will see them all, just pick the one(s) you want.
6. Your output should be an AC3 or DTS file, or something similar to that. If you completed the ffmpeg installation correctly, these should now be openable in Audacity.
7. You will get a multirack file in Audacity with 6 mono channels.
8. Order of channels should be L / R / C / S / LS / RS.

Center channel will generally give you access to voices. It's not always 100% clean access to the voice, but generally the music and stuff is turned way down on this channel making the voice sample easier to work with. Sampling from L/R allows you to get music samples without the voice overs. The Sub and Surround channels offer some opportunity for getting effects samples from movies.

Not sure if this helps anybody but cheers if it does. I wrote it up for a blog, but figured I would share here as well!
Well, DVDAE works as well. But it only works on DVD's specifically, and costs money.

I was aiming for something free, and flexible. (While the write up was MKV focused, the same concepts can be applied to basically any video type. There are open source tools for muxing/demuxing basically any format imaginable).


VIP Junglist
u can use video files as long as they have 5:1 sound on them as well with dvd audio extraction.

Much simpler ways to do things these days