How to use Cool Edit Pro/Adobe Audition to take obscure songs you like and make them into CDs you can take to karaoke!

---

My friend Tim came over today to make some karaoke versions of his pavorite songz. I've done this for awhile now, so here is a quick tutorial on how to do to it using Cool Edit Pro (now aka Adobe Audition, but old version of Cool Edit Pro have all the functionality you need.)

First, rip the track off the CD. Usign MP3s works, but filtering will be less effective. Save the original CD audio to a file, like "song.wav".

Second, use the Channel Mixer (under Amplitude) on the original audio and set Left/Left to 100%, Left/Right to 0%, Right/Left to 0%, and Right/Right to 100%. Make sure the 'Invert' checkboxes are unchecked. Save this to a file called "song-uncenter.wav". If you listen to this "uncenter" track and don't hear a lot of vocals (reverb echoes, etc) then it is probably a good candidate for karaoke remixing. If there are lots of vocals still heard, your job will be much more difficult.

Next, use Pan/Expand (under Amplitude) on the original audio. The Center Channel Pan should be dead-center. The Stereo Expand should be 0 -- a flat line along the bottom of graph. Do this and save it as "song-center.wav".

Next, find a part of the chorus or verse in song-center that has the music but no vocals. (This can be difficult with, say, Speed Metal.) Highlight as much of the segment as you can without vocals and press Control+C to copy it to the paste buffer. Make a new file (Control+N with defaults) and paste this segment into the new file. Do this again for the verse or chorus only -- don't do this for the whole song. Once you've pasted all the non-vocal segments in the chorus or verse into your new file, listen to it and make sure there aren't any vocals in it.

Use Noise Reduction (process -- under Restoration) on your new file of non-vocal segments. For Snapshots in Profile, put in between 250-500 per second of non-vocal segments. Note that the more snapshots it uses, the slower it is to capture the profile. A *lot* slower. Once the profile is captured, click Close. Go back to song-center and highlight pieces that have vocals from the section you did you noise profile on (don't apply this noise filtering to parts you didn't specificall capture -- stick with the chorus or the verse) -- leave out the non-vocal pieces. Use Noise Reduction again, set the FFT size to 24000, select Keep Only Noise, set the Precision Factor to 11, Reduce by 65dB, and keep the Smoothing and Transition settings at 0. Click OK.

You'll notice you can still hear the vocals pretty well, but they sound weird. Now we'll use the FFT Filter (under Filters) to get rid of as much as we can while leaving as much of the 'groove' of the center channel. Highlight the entire file for the song-center. Open the FFT Filter and: set the view to Passive, check Log Scale and choose an FFT Size of 16384 (for now). Under Presets, choose 'Only the Subwoofer'. This gives you a nice template for how to cut out the vocal frequency range. I suggest starting by cutting out 440hz-3520hz. Your graph should look like this:

|
|
|-----------------------+                +--------------------| 100%
|                       |                |                    | 80
|                       |                |                    | 60
|                       |                |                    | 40
|                       |                |                    | 20
|                       +----------------+                    | 0
                       440              3520

This will definitely make your song sound odd when you only listen to the center. Don't worry -- it will sound better when we mix the uncenter back in.

Just to illustrate this quickly, go to the Multitrack view by pressing F12. Click in the empty space next to Track 1 and select Insert. Under the Insert pulldown menu, select the song-center.wav. Note that when you edit song-center.wav in the Wave Editor, it will automatically be updated in the Multitrack View -- you won't have to re-insert it. Then, click in the empty space next to Track 2 (or press the down arrow to bring the focus to track 2) and Insert the song-uncenter.wav. At the bottom of the multitrack view, click play. It will mix both tracks together in real-time and play them. As you'll probably hear, the full fidelity of the uncenter makes up for the loss of fidelity in the center.

That's it for part 1. By now, you should be able to basically finish a track. If you have questions, leave a comment on this post and I'll answer them in part 2.

PS: I am such a fuqn nerd, thanks for reading.

Discussion: showing all 0 messages

(No messages)

>

Post a new message:

---