Hi all,
I’ve used the vocal remover effect in Audacity 1.3 beta version with no problem (I mostly use it for removing people talking from the background instrumental music), but today somehow when I tried the vocal remover (still with 1.3 beta) I can still hear the vocal, though it sounded like it’s far away and more static-y. So I updated it to the current version but still with the same problem.
Does anyone know what happened?
A example of what I’m talking about here
The talking starts at around 1:48
You changed music. Vocal Removal fails a lot more often than it works. You lucked out big time if it worked for you multiple times. It doesn’t work on mono shows, highly compressed shows, shows with the singer not in the middle and shows with stereo effects added to the voice.
That’s pretty much every show out there.
People keep pointing to the guy on YouTube who did it the hard way with the manual tools and the terrific results he got. We tell them if they used the identical song he did, you can get terrific results, too.
I have a clip here somewhere that had very deep echo effects and Vocal Removal was a complete washout.
Thanks for the reply, but it worked for me all the time, and I have a successful product as well, the vocals were completely wiped out. For some reason, it just won’t do it again to the exact music…
You didn’t say that earlier. Are you in fact using the Vocal Remover on a clip that used to work perfectly?
I can tell you that the clip you posted has very serious phasing problems. The show changes direction abruptly as I wave my head in front of the speakers.
Let me grab the clip… Yes, the upper and lower channels are exactly out of phase. Vocal Remover doesn’t do that. It’s output is a single mono show. Are you sure you used the Audacity Vocal Remover?
Did you do anything to change the clip between the time it worked and now?
Compression is a real problem with Vocal Remover. As the compression passes go up, the relationship between left and right gets fuzzy and that’s deadly for channel phase cancellation. Any time you enter and exit Audacity, it has to recompress the work. It’s not an MP3 editor.
Also, the new Vocal Remover can be tuned – it has frequency range limits and sliders.
Just on a SWAG (scientific wild-ass guess). Try to play and vocal remove this clip.
Only the third segment should vanish. The other four should remain. It’s 39 seconds.
You also didn’t say which Windows you had. Our Win7 machines have extensive special effects that you can add to the show on playback. I got burned once when one of the other techs left “Concert Hall” running by accident and I was doing sound tests…
Hey, thanks for the reply.
I don’t really understand what you mean by “Only the third segment should vanish. The other four should remain. It’s 39 seconds.” but I’ve done the vocal remover and I’m posting it back up for you to check. http://kiwi6.com/file/qidgbi7521
I actually did switch from Simple (entire spectrum) to the other two options in the vocal removal option to hear the differences between them… did that cause this?
You don’t need to send it back. The object of the test is, first, to make sure your speakers and sound system are wired right. This is the web site associated with the clip.
If you listen to all 39 seconds just straight playback, the third segment should sound flat and plain (and mono) and the fourth segment should sound hollow or like it’s coming from behind you. If those two are reversed, then you have a serious sound management problem somewhere (traditionally, your speakers are wired wrong).
If you have trouble hearing the differences, point your two speakers directly toward each other so they touch. Segment three should be much louder than segment four.
The third segment should vanish completely if you vocal remove it. Did it? The left and right in segment three are as perfectly matched as the digital system can make it. If you have any residual voice left, then there’s something wrong. If you amplify the removed segment and turn your speakers up, you might hear a slight suggestion of the voice left. This is the accuracy limit of 48000, 16-bit audio