Vocal removing issue

[george85]

Hi there.
I downloaded the Audacity 1.3.13 Beta and Lame_v3.98.3_for_Audacity_on_Windows
I found your tutorial about removing the Vocals from the song:
http://www.youtube.com/watch?v=PqXiKYG3J7M

I did what it was told in the tutorial - split into stereo track, then effect->invert, then set back to Mono, both trucks.
The results is nice, but still some words can be heard.

In youtube tutorial, they say that I need to use the negative value of effect->amplify. But they did not say on which tracks do I need to do this? Select the first track and then do the Amplify, or the second one? Or select both? I tried all three case, and for some reason, the words are coming back when I do the Amplify command, so its is getting even worse.

Maybe I need to save the file as Mp3 and then, import it again in the Audacity, and do the effect->amplify?

Any help please.

Here is the song with vocals:
http://www.gamefront.com/files/20729759/SONG.mp3

Thank you
George

[steve]

I found your tutorial about removing the Vocals from the song:

That's not "our" tutorial - it's just one that someone made.

I did what it was told in the tutorial - split into stereo track, then effect->invert, then set back to Mono, both trucks.

Audacity 1.3.13 has an effect that does this all in one go. It's called "Vocal Remover (for center-panned vocals)" and you'll find it at the bottom of the Effect menu.
It does the same as what you were doing manually, just less mouse clicks.

The results is nice, but still some words can be heard.

That's probably as good as you'll get. See this page on the Audacity wiki: http://wiki.audacityteam.org/wiki/Vocal_Removal

[george85]

Thank you for the reply Steve.

I know you receive tons and tons of topics like this. Sorry for that.
And I am apologizing for naming the youtube video as yours. I did not know. Sorry

I also used the Extra boy plugin (both modes) in Audacity, and still the same result.
I also used the Power Audio Editor, and again the same result.
AnalogX plugin for Winamp - again the same result.


So like you said, that`s the far I can go. But what about Amplify:

I need to use the negative value of effect->amplify. But they did not say on which tracks do I need to do this? Select the first track and then do the Amplify, or the second one? Or select both? I tried all three case, and for some reason, the words are coming back when I do the Amplify command, so its is getting even worse.

Maybe I need to save the file as Mp3 and then, import it again in the Audacity, and do the effect->amplify?

?

[steve]

apology not needed - we're all Audacity users here and it's just good if we can sometimes help each other out

The bit about "Amplify" is (how do I say this nicely...) rubbish
All that "Amplify with a negative value" will do is to make the entire thing quieter.

IF the vocals that you want to remove are slightly off centre, for example if the vocal is slightly to the left, then the vocal removal process will be a bit better if you negatively amplify the left channel before doing the vocal removal - the idea here is to adjust the left / right levels so that the vocal to be removed is dead centre of the stereo mix. Nice in theory, but not very practical in practice. In practice, if the vocal to be removed is not dead centre, then you are unlikely to be able to judge how much to amplify which track to make it dead centre. In some cases you can try amplifying one channel just a little and see if that makes the vocal removal better or worse, but generally it's not worth the effort.

There's a couple of tricks that he missed in that video - rather than splitting the stereo track and then converting the tracks to mono, there's an option in the track drop down menu of Audacity 1.3.13 to "Split to Mono". If he used "Split to Mono" rather than "Split Stereo Track" then the two track would already be set to mono so he would not need to convert them.

Also, his original track level was excessively high and you can see the red "clip indicator" as the green playback meters hit the top of the scale. It is advisable to always keep your levels below 0 dB so as to avoid distortion.

"Saving as MP3 and re-importing" does nothing apart from lowering the sound quality. Encoding as MP3 always reduces the sound quality a little - the data compression discards some audio information so that it can get the file size really small, but that lost data is not recoverable. Encoding as MP3 should only be done as the last step in a production process - until then, try to keep everything in high quality uncompressed formats. Where possible, start with WAV files and only encode to MP3 on the final Export.

[george85]

Steve I owe you a beer for all this. Honestly. Just about this you said:

"IF the vocals that you want to remove are slightly off centre, for example if the vocal is slightly to the left, then the vocal removal process will be a bit better if you negatively amplify the left channel before doing the vocal removal - the idea here is to adjust the left / right levels so that the vocal to be removed is dead centre of the stereo mix."

I do not understand. You have just told me that is more easier to use the "Vocal Remover". Well in that case I can not amplify the left or right channel, as process of separating the stereo track is one of the steps "Vocal Remover" automatically does, doesn`t it? Or am I wrong?

I do not know if the vocals are right, or left.
Can I ask you for a favor, here is the song, can you take a look at it:
http://www.gamefront.com/files/20729759/SONG.mp3

Thank you Steve, I owe you big time.

[steve]

You have just told me that is more easier to use the "Vocal Remover"

Yes it is - 3 clicks and your done.

Well in that case I can not amplify the left or right channel, as process of separating the stereo track is one of the steps "Vocal Remover" automatically does,

I also said that it's probably not worth bothering to try and re-centre an off centre vocal (because it's likely to be a lot of effort for little or no improvement).
However, if you really did want to try to centre an off-canter vocal so as to improve the vocal removal, you would do so before applying vocal removal and in this case it would be easier to use the manual (split and invert) method rather than the plug-in. Here's how you would do it:
1) Split the stereo track using "Split to Mono" from the drop down menu.
2) Select one of the tracks (say the lower of the two) and apply "Invert" from the Effect menu

If you now press "Play" you will hear if there is any vocal removal occurring. If you are not hearing some degree of success at this stage, give up, it's not going to work.

An interesting "feature" is that if the original stereo track was recorded in mono, you will now hear "silence" (or more accurately, you won't hear)

Introducing a fairly new feature in Audacity 1.3.x - the Mixer Board:
3) Open the Mixer Board (View menu).
You will see two large vertical sliders - the one on the left is for the first track and the one on the right is for the 2nd track. These sliders duplicate the "gain" sliders on the left end of the tracks. Moving the slider up will increase the playback level (the track gain) and lowering the slider will reduce the track playback level.

4) Adjust the sliders to achieve the best vocal removal. Take care not to boost the playback levels too much or the sound will distort. Keep an eye on the green playback meter and ensure that it stays below 0 dB. Note that you can move these slider while the track is playing. (this is one of the very few things that you can do in Audacity while Audacity is playing).

5) When you're happy about the "mix", (you can close the Mixer Board now), select both tracks (Ctrl+A will select All tracks) and from the Tracks menu select "Mix and Render". This will mix the two tracks down into a single mono track - hopefully with most of the vocal missing.

I do not know if the vocals are right, or left.
Can I ask you for a favor, here is the song, can you take a look at it:

It's in the middle.
Use the "Vocal Removal" effect from the Effect menu.

[kozikowski]

Having the vocal off-centre left to right isn't the only problem you may have. That's a significant one, of course, but if the voice has had any stereo effects added to it (room echo, artificial separation) then the effects will stay in the show even if the voice is gone. This gives the "ghost voices" people complain about.

The tool is subtitled "Center Pan Removal" because that's what it really does. Everything in the middle of the show goes, frequently taking the drums and bass with it.

And if you haven't caught on by now, you don't get stereo music at the end. The tool gives you a mono show.

This is a stereo test track that I recorded a while back. You should find that the third cut, "my voice in the center," vanishes when you put it through the Vocal Removal tool. But none of the other cuts. Cuts one and two fail because I'm not speaking in the "middle" and cut four fails because of simulated recording damage.

http://www.kozco.com/tech/LRMonoPhaset4.wav

I don't remember if we ever asked you if you have a mono show. None of these tools work on a mono show -- at all. You can have two tracks with blue waves on the time line and still have a mono show by having exactly the same music and waves on both. Look at the green playback sound meters. Are they exactly the same during playback of the show, or are they different like this...

http://kozco.com/tech/audacity/Audacity1_playback.jpg

Koz

[george85]

Thank you for the repliesa again.

@kozikowski: These two waves are the same before I do the Effect->Invert. But After I do this, the lower one changes.

@steve: Mixer board did not worked. I tried playing around with both left and right sliders, and if I change both left (-0 dB) and right (+0dB) values the lyrics are starting to appear again.

One more thing: when you said this:

However, if you really did want to try to centre an off-canter vocal so as to improve the vocal removal, you would do so before applying vocal removal and in this case it would be easier to use the manual (split and invert) method rather than the plug-in.

I got a little bit confused, as we mentioned "Amplify" and now "Mixer board". So In case of Amplify, I would do this:

1) split stereo track,
2) negatively amplify the left (upper) track
2) select the right (lower) channel, and Effect->Invert,
3) then set back to Mono, both channels.
4) File->Export.

Is this what you were saying?

[steve]

@steve: Mixer board did not worked. I tried playing around with both left and right sliders, and if I change both left (-0 dB) and right (+0dB) values the lyrics are starting to appear again.

Exactly as expected.
The vocal on your original track is dead centre, so when one channel is inverted the vocal in left channel almost perfectly cancels out the voice in the right channel when both of the sliders are set at 0 dB.
If you move either of the sliders then in effect it is pushing the vocal to either the right or left of centre and they no longer cancel out.

I got a little bit confused, as we mentioned "Amplify" and now "Mixer board".

The "Amplify" effect changes the amplitude of the waveform in a track (makes it vertically bigger or smaller). The bigger the waveform the louder it will be, the smaller the waveform the quieter it will be. The actual audio data is changed when you apply the Amplify effect.

The "Track Gain" slider, does a similar thing, but without changing the actual audio data. What it does is to change how loud the track is played. It's a kind of track volume control. When you increase the gain slider (move it to the right) then the track will play louder. When you decrease the slider (move it to the left) then the track will play more quietly. This does not change the audio data at all, it just changes how loudly that audio data is played.

The big sliders in the Mixer Board are exactly the same as the track gain sliders. In fact, when you move one of the sliders on the Mixer Board you can see that the track gain slider also moves. The big sliders on the mixer board are a bit easier to use than the little track gain sliders (because they're bigger).

So In case of Amplify, I would do this:

1) split stereo track,
2) negatively amplify the left (upper) track
2) select the right (lower) channel, and Effect->Invert,
3) then set back to Mono, both channels.
4) File->Export.

That's what you would do if the voice was left of centre. The problem is that you would not know how much to negatively amplify the left (upper) track.

[george85]

So In case of Amplify, I would do this:

1) split stereo track,
2) negatively amplify the left (upper) track
2) select the right (lower) channel, and Effect->Invert,
3) then set back to Mono, both channels.
4) File->Export.

That's what you would do if the voice was left of centre. The problem is that you would not know how much to negatively amplify the left (upper) track.

So even in case I would know how much to negatively amplify the left (upper) track, I would still not be able to use this method, as the vocal on my original track is dead centre, not left of center, right? Or to be precise, I can not do anything else to remove the lyrics from my song. That`s it?