How much better is a CD rip of a song for vocal removal?

Long story short; I just ordered a sampler CD off of Amazon because it has a specific song I want to eliminate the vocals from. Before you say it won’t work, I personally bought the MP3 download of the same song, tried it, and it removed every trace of the lead singer, but the drums were taken out, and the quality decreased. I think the fact the MP3 itself is merely 174 kbps is why the quality decreased quite a bit. It did get me thinking, though; Would I get a better result if I ripped the song from the CD into a WAV file with CDex, and then tried it on the ripped version? Thanks all! :slight_smile:

The drums vanished because they were in the stereo middle of the show along with the lead singer. Vocal Removal used to be called “Center Pan Remover” because that’s what it did. It removed everything in the left-right center of the show.

It still does, but Vocal Mover in Audacity 2.1.0 and 2.1.1 has additional tools to help with the effect of removing too much stuff.

Koz

Oh, I know how it works. I was just wondering if the result would be any better with the CD ripped version. i know the drums would still disappear, but would the quality be a bit better as opposed to the MP3?

It will always be better, but the question is how much better and if the improvement is audible. That depends on who made the MP3, the quality setting and how many conversions it’s been through in the middle.

Koz

but would the quality be a bit better as opposed to the MP3?

That’s a tough one… You won’t know unless you try it both ways and once you have the CD there’s no point in trying it both ways…

Will you feel bad if you spend money on the CD and there’s no difference, or is the cost of the CD insignificant to you?

The rule is use lossless formats for audio production (whenever possible). These formats are for audio distribution (if that’s the format you or your listeners/customers want). If you were doing a commercial production, I’d say absolutely use the CD, or the highest quality source available. (But, I’m pretty sure you’re not doing a commercial production,:wink: )

If you start with an MP3 and your final production is MP3 compressed, there will be additional quality loss from the 2nd lossy compression (which may or may not be audible).

The center-channel removal might tend to bring-out MP3 compression artifacts (although I wouldn’t expect that).

On the other hand, a good quality MP3 (such as from Amazon) is usually audibly identical to the CD (in a proper scientific, blind, listening test).

And, the “damage” done by center-channel removal is going to be worse than the “damage” done by MP3 compression, so starting with a high-quality original may not make a difference.

In my experience, vocal removal is a novelty and rarely useful for anything “serious”. So personally, I’d just go-ahead and use the MP3.

The “help screen” for the legacy Vocal Remover plugin says lossless files are better than MP3 or compressed formats.

Gale

Well, that’s just a statement - it lacks some reference material.
Perhaps an opportunity for a master thesis…

Here are my observations:

It does essentially not matter if you’re remove the center from a CD or an Mp3, the residual is about the same (faint lead vocals in the background).

However, Mp3 (at low bit rates) removes a lot of quiet sounds that are normally masked by the louder sounds of the audio. This works fine for a stereo track as whole but not if the center is removed or isolated. The consequence is that you’ll hear all the musical artefacts that would normally be masked.
This effect is further amplified by the way Mp3 encodes stereo. The default kind is “Joined Stereo”. The width and pan position of a track can be quantized more aggressively than the actual samples in a channel separately. Thus, the pan position could be reduced from 65536 possibilities to only 256 (You won’t notice this quantization, I bet).
This will be audible after vocal removal (and more so after isolation) as “pumping” because the center jumps around due to the quantization rounding.


What’s more important to keep the drums:

  • set the low cut as high as possible, just below the lowest sung note.
  • apply the effect only where something has to be removed (use Ctrl+R to repeat the effect).
  • use different filter settings if e.g. the chorus is much higher than the verse.
  • Strong drum beats can be isolated beforehand; I use a simple drum filter algorithm for that.

Robert

Once the CD comes in the mail, and I try it, I’ll update you all.

UPDATE: The vocal remover worked 100x better than the crappy MP3 did. No quality loss whatsoever, besides the drums being absent, it could be used for karaoke without a doubt.

100x better than the crappy MP3 did.

Again, it’s probably not “one” MP3. It’s probably been through multiple MP3 compressions. I bet if you play the CD and play the MP3, they don’t sound the same. They’re supposed to. A properly made MP3 file is indistinguishable from the original performance. So you don’t just have an MP3. You have a broken and badly damaged sound file.

Koz

Understandable. Anywho, here is a video of how good it really did with the CD version in case you wondered:

https://www.youtube.com/watch?v=dx7If7AEJ_o