WEBVTT
Kind: captions
Language: en

00:00:00.660 --> 00:00:02.680
Hello, I'm Wingbeat Pony.

00:00:02.780 --> 00:00:08.750
Today, I'll be showing you how I do the Vocal-free MLP project, plus a few other ways to remove vocals.

00:00:09.000 --> 00:00:13.160
We'll be working in Audacity, so open it up and bring in your audio.

00:00:13.510 --> 00:00:20.570
Now, if you ripped the audio out of an .AVI or off of YouTube, you'll probably be looking at a stereo track like this.

00:00:21.600 --> 00:00:26.120
Applebloom: Today's the day, cutie mark crusaders! I can just feel it!

00:00:26.220 --> 00:00:34.010
Audacity has a one-step process to remove vocals from a stereo track. Look in Effects, and you should find "Remove Center-Panned Vocals."

00:00:34.450 --> 00:00:36.560
The result sounds something like this.

00:00:42.270 --> 00:00:48.570
Terrible, right? Why's it suddenly mono? Well, let me show you what just happened by doing it manually.

00:00:49.170 --> 00:00:57.150
Vocals tend to be recorded in mono, while the music is in stereo, so only the vocals are exactly the same volume in both the left and right channels.

00:00:57.770 --> 00:01:02.800
So, if one side gets subtracted from the other, only the mono part gets cancelled out.

00:01:03.490 --> 00:01:08.600
That's why I've split the stereo track into two mono tracks: So they can be subtracted.

00:01:09.330 --> 00:01:17.750
To do that, one side has to get inverted. This flips the waveform upside down, so adding them together results in zero. Make sense?

00:01:18.630 --> 00:01:21.580
If we listen to the result, it's exactly the same as before.

00:01:27.730 --> 00:01:33.290
So how do I do it in stereo? How do I get voices removed when they're not perfectly centered?

00:01:33.970 --> 00:01:41.740
To start with, I don't use a 2-channel stereo track. Instead, I use the Dolby 5.1 surround audio as my source.

00:01:42.230 --> 00:01:45.130
You can get that by buying the episodes on iTunes.

00:01:45.520 --> 00:01:48.530
BUY THE EPISODES ON ITUNES.

00:01:49.370 --> 00:01:56.180
Then demux the audio, something I won't go over in this tutorial, but you should be able to find out how elsewhere on the Internet.

00:01:56.920 --> 00:02:02.990
Anyway, open it up, and we get this. Six audio tracks, all helpfully unlabelled.

00:02:03.690 --> 00:02:12.030
See track 3? It's a lot louder and busier than the other tracks, and you can be sure that's where the vocals are. Mute it and hit play to make sure.

00:02:12.840 --> 00:02:16.160
Now, let's re-label these tracks and put them in the proper channels.

00:02:16.850 --> 00:02:21.780
The first two are left and right, followed by center and bass (which stay mono).

00:02:22.280 --> 00:02:26.390
Then there's left and right surround. Just put them on left and right.

00:02:27.500 --> 00:02:31.320
Now, I could just mute the center channel, but I'll want it for editing later.

00:02:31.510 --> 00:02:32.590
So invert it...

00:02:32.970 --> 00:02:34.750
...and duplicate the track.

00:02:35.470 --> 00:02:44.720
Rename the tracks as "Left Vox" and "Right Vox." Also, take the envelope tool (that's the blue one) and drag the envelopes down to zero to mute them.

00:02:46.020 --> 00:02:48.730
90% of our work is done. Give it a listen.

00:02:54.640 --> 00:03:02.790
Now, occasionally a character will say something and the audio won't be perfectly in the center, for instance when they're walking accross the screen like Zecora here.

00:03:04.880 --> 00:03:08.650
Zecora: ...I have just the trick, that will fix you up quite quick.

00:03:09.250 --> 00:03:17.370
Thankfully, the audio is rarely panned all the way to one side or the other. That means there's still a little audio left in the center we can use to cancel it out.

00:03:18.230 --> 00:03:23.060
Drag the envelope for the appropriate channel back out, and you'll hear the voices get a little quieter.

00:03:24.530 --> 00:03:28.470
[Quieter] Zecora: ...I have just the trick... fix you up quite quick.

00:03:29.210 --> 00:03:36.010
It's not perfect, and that's because the panning is changing through this line. Play around with the envelope until you can't hear it at all.

00:03:36.800 --> 00:03:38.170
This is the end result...

00:03:44.750 --> 00:03:54.240
That's it! Now, there's still a little trace of the line due to the fact that I can't match the envelope perfectly, but it's a far cry from not being able to remove it at all.

00:03:54.960 --> 00:04:01.830
There are other artifacts, like reverb in indoor scenes and BOOMING, ECHOING VOICES that I really can't remove.

00:04:02.410 --> 00:04:07.930
The reason for that is that they're offset a little in the stereo mix, so they can't be lined up with the center channel.

00:04:08.560 --> 00:04:13.890
It's bad news for me, but evidence of stellar sound mixing on behalf of the creators of the show.

00:04:14.440 --> 00:04:18.410
Thanks for watching, and I hope you stick around to watch more Vocal-free MLP!