Extracting (De-Multiplexing) Sound From *.MPG Files

There are two possible ways to extract audio from an MPEG file. The first is by de-multiplexing, or breaking the video into its component parts- a video stream and an audio stream. This is the method of choice for videos downloaded from the internet, which probably have highly compressed sound to begin with--you don't want to compress it further. The second method is simply re-encoding the composite video to output a solo audio stream. This method is good if your audio stream is high quality and you want to compress the audio to save hard drive space.

Here's how to de-multiplex an *.MPG file:

1. Open TMPGEnc, and a wizard box will pop up. Close it. Click on [File / MPEG Tools] and you'll see a window similar to that below:

2. Click on the De-multiplex tab and you'll see a window like the one above. This window will be blank when you first open it. Click on Browse and find your video file.

3. You want to de-multiplex the audio portion, so double click on the highlighted audio_stream 0x00 line and a dialog box will pop up asking where you want to save the audio output. In most *.MPG files, the audio is stored within the video in a *.MP2 format, so naturally that's what's going to get outputted--not an *.MP3. If you absolutely need *.MP3 format for a portable player or something then use Winamp to convert it.
Side note- if you feel so inclined, you can also double click the video stream to save only the video data without sound. This is useful for re-dubbing.

Anyway, once you choose where you want to save the file, you're done.

Now I'm going to go over how to re-encode a solo audio stream:

Since I'm too lazy to make all new graphics with minor changes for the purpose of this tutorial, I'm just going to use the screenshots from the demuxing tutorial. Ignore any red highlights because they won't pertain to this exercise.

1. We're looking at the bottom half of the screen. Next to the Video source textbox is a Browse button. Click it and find your video file. The Audio source box will fill itself in using the same path as the video source, naturally because you want the audio stream from the video file you specified.

2. Below the Audio and Video source boxes, is one that says Output file name. Click the Browse button and point to where you want to save the output, and the filename you want to save your output as.

3. See the frame to the right of the Audio, Video, and Output textboxes that's labeled Stream type? Make sure you select ES (Audio only). It means you want an Elementary Stream, Audio portion only. Given that, you should be able to figure out what the rest of the options do, and if you're dense then there are ToolTips that explain them further.

4. Now we need to find the maximum audio bitrate we can save the audio as. If you skip this step, you may end up saving a high-quality copy of a poor-quality audio stream, which will do nothing but eat up space on your hard drive. So we want to set our desired audio quality (bitrate) relative to the quality of the audio contained within the actual video file. To check the bitrate within TMPGEnc, go to [File / MPEG Tools]. A window like the one below will pop up:

5. Click on the De-multiplex tab. Here's where you ignore the highlighting. Click on Browse to find your file, then look down the Information column to find the values that correspond with the audio stream (labelled audio_stream 0x00). The audio stream of the file I used is 44100Hz 96kbps *.MP2. Write down whatever it says for yours, then click Cancel.

6. Go back to the main screen, and under where you selected ES (Audio only) is a button marked Settings. Click it and you'll see something like below.

7. Now you can pay attention to the highlighting. What I've drawn a box around are the parameters that we just looked up, that we now want to configure. If you've come this far you probably want to set these values to something LESS than what we looked up, otherwise you would have been better off doing a simple de-multiplex.

Set the sampling frequency to whatever the value we looked up was (in this case 44100Hz). Set the channel mode to Stereo or Mono depending on whether the video is Stereo or Mono, or whether or not you care. If you're concerned about filesize select Mono. The bitrate, now, is the part where we most affect the output file size. As you can see, my screenshot shows I'm about to save in 44100Hz 128kbps Mono when the video file I'm working with is only 44100Hz 96kbps Mono. So if I continued, I'd end up with a bloated output file that still only sounds like a 44100Hz 96kbps Mono recording. You should change your bitrate to be something lower than whatever the source was (in my case I should save as something less than 96kbps).

You can also tool around with the audio editor, to crop the sound or whatever. You can do it now, or later with an external sound editor. Click OK when you're done.

Now you're back at the main screen. It's not in my screenshot since a menu is covering it, but there's a big button in the top left corner marked Start. Click it and TMPGEnc will re-encode a solo audio stream to the location you specified earlier. You're done!

Kopywrong (k) 2005
All Rights Reversed. Reprint what you like.