Vocal dampening?

Um... just a dinky text thing for those interested in how vocal dampening is done. These post are excerpts from posts by AX Karaoke Head Tech, Hong Kwon, and AMV master, Justin.

Msg 421 from AX Karaoke Club on Yahoo:
Vocal dampening as used by typical karaoke systems is really just a cheap equalizer that cuts middle frequencies (which is the spectrum where voice mostly falls). This can be attained by using a good parametric EQ and filtering out certain mids. However a good parametric EQ is overkill for this situation (unless you really want a high degree of control over the audio, which is usually not necessary for anime karaoke ;-)

What we do is we have a karaoke signal processor, which is a box that sits between the music source and the PA and does the voice filtering and also pitch select (higher/lower). I got mine for about $200; you can find these kind of things at Asian karaoke stores under the name "karaoke processor" or "karaoke mixer". Of course if you already have a karaoke cd/dvd/ld player then it probably has this feature built in. The advantage of the outboard karaoke signal processor is that you can use it with any audio source, including computers with mp3s (what we use). Just plug the audio source into the unit, plug the output from the unit into the PA, and you're good to go. Our unit also has reverb for that semi-authentic karaoke box feel.

The other thing we use (but not as much) is a Winamp plugin called DeFX which does the same thing for mp3s. The drawback to this is that it doesn't work with CD tracks, only mp3s, and it also doesn't with work mp3s that are VBR (variable bit rate) encoded. If you have constant 128 bitrate (or higher) mp3s, then DeFX will work for you. VBR encoded mp3s, when played through the plugin, sound squiggly and weird due to the processing algorithm not being prepared for changes in the bitrate.

One other possibility to pursue is the Sega Saturn (!). If you put in an audio cd, there is actually a karaoke option (vocal damper) built into the cd player UI. :-) This was used at AX96 karaoke...

Response from Justin on ezboard
Good Karaoke algorithms don't apply high/low pass filters, they use Stereo Separation to determine which parts are the voices.

An example: If you have a tuner in your living room or something with Dolby Pro-Logic Surround, and you have a surround sound speaker setup, play your music into it and unplug the center speaker. Most of the vocals should now be gone.

What happens is that when a song is recorded, generally the vocals are given the exact same intensity on both the left and right channels. This means that what you can do is take the Difference of the two channels (make the left channel = left-right and make the right channel = right-left) and then you have the voice removed, for the most part (except for echoes).

This is usually how good karaoke machines work, and how I would suspect yours works. The way you can tell is if you still get Backup Singers when you turn on the kararoke mode. If you do, then obviously it's not simply using an equalizer, because if it was, then it would filter out ALL human voice, not just the main one.

I've actually gone and created a Karaoke version of the song I'm doing myself on my computer and have burned it to CD (I own the original of course) because this way I have complete control over what the final product will sound like. I've obtained Karaoke versions of the other two songs (which actually exist) but one of mine does not have a Karaoke version in existance, anywhere, so I've made one myself (which sounds pretty good, you do lose some high percussion though).

I hope this has been educational =)