Creating Backing Tracks Using MP3 Vocals and Instrumentation

Hi Conrad, good to see you back. Inside SONAR this task can be a bit cumbersome. R-Mix (which came with X2) is probably simplest, but still cumbersome as you must go track-by-track. I realize you are thinking track-by-track at the moment, but let me explain. I understand your purpose, but also realize that digital information is embedded into most commercial tracks, so when done, there is a good chance you cannot post such editing on web sites (SoundCloud, YouTube, etc.). I wanted to make sure that point was clear before I start.
 
(Batch) removing vocals -
Easiest method would be to get a trial copy of Adobe Audition CC. It is good for 30 days, (used to be) fully functional, and can do batch processing (i.e. you can remove vocals from every mp3 you have on your computer in probably 20 minutes). Removing vocals with its default setting will leave some residual, low-level vocal with a flanging effect, and can also be destructive (again, flanging effect) to vocal-range instruments that are center-panned.
 
In Audition:

1. Go to Edit->Batch Process, this will open a window.
2. Click Open (upper right of that window), and can import up to 100 mp3s at a time typically. Click the first and then Shift-Click the last, and you will select everything in between. Let them load, and you may need to adjust count based on your computer's performance.
3. Next, in the lower right, click "Batch Process..." and that will open another window.
4. At the top, for "Apply Favorites" choose "Remove Vocals" from that dropdown.
5. Choose either a prefix or suffix for the tracks, such as "NoVocal_" then [song name] will become NoVocal_[song name]
6. For location, choose "Same as file location" be sure "Overwrite Existing files" is unchecked (just in case).
7. Choose output format you want.
8. For the bottom three check boxes, check "Clear Batch Process Upon Completion," check "Close Files Upon Completion," and uncheck "Include Markers and Other Metadata"
9. Click "Okay" and it will process one song/thread and close/remove them from the process window as each completes.
10. When done, you will have another set of songs in your directory, all with "NoVocal_" at the beginning.

This option would also allow for you to have a set of songs without vocals that you "may one day" want, long after Audtion's trial ends.
 
Adding your vocal to the track -
The above process will leave residual vocal with flanging that you can sing over (drown out) by using a side-chained compressor. Edit: Oops, forgot to mention, keyword here is "ducking."
 
In SONAR:

1. Create a project and insert both the original and no vocal copy into tracks 1 & 2. It may be easiest to record your vocals when hearing the original.
2. Sing vocals (to track of your choice), comp/edit as normal.
3. To sidechain a compressor, on the no vocal track, in the ProChannel, insert the "PC4K Channel-Type Compressor"
4. On your vocal track, right click the left pane->Insert Send->PC4K Exp/Gate
5. On the PC module (no vocal track), enable "S.Chain" so the compressor is responding to the side chain. For settings you typically should not need a lot, only a couple dB compression will work wonders. You may need to look up more info on usage if necessary.
6. Can mixdown the no vocal track and your vocal when suited to taste.

 
That is a high-level overview, but again bear in mind that some tracks may not be "very usable" with Audition's vocal removal, and even if so, you may still be unable to post them afterwards. Even if they do post, be sure to mention the vocals are yours and the rest are not just to be on the safe side with things.

Hey Conrad. Creating a backing track from scratch is certainly doable, but even with proficiency at it is not quick. MIDI files can be hit or miss at times, but tend to be the best starting point. Even then they will often require some editing, so quelling the urge to make a "Xerox" of the original will save a lot of undo "pain and suffering" in the process. From a learning perspective, there is a lot to be had in such an endeavor. Just start with easier pieces until you get accustomed to a workflow that suits you.
 
One thing that came to mind with my above comment is when using mp3s you may need to nudge one of the two original tracks to line up an easily spotted transient (like the first snare hit). Mp3 files sometimes load with a sample offset for the start time, so zooming into a nice transient will let you mate them up sample for sample (if zoomed in closely, you can drag one clip to mate them, but you cannot drag left if the clips are butted to the zero time (left edge of the track view), so drag the "early hit" clip to the right). That way you can work to the track you choose and mute back and forth between them without throwing your vocal off. FYI - I tend to start clips at measure 2 so 1) I can drag left if needed and 2) if AudioSnap is used I have room on the left edge to work with.
 
Nice choice of car! Big, red, and a V8... all good
 
Edit: Another thing is I have Audition from Adobe CS5.5, but doubt the workflow I gave you above has changed much. I wanted to point that out just in case some buttons/menus moved. Audition is my go-to wav editor, so if you like that program, there are still OEM copies of CS5.5/CS6.0 to be had. For what you are doing I would not recommend subscribing to the CC version. With the CC version, if you stop subscribing it is gone, but the CS5.5/CS6.0 versions are "one time buys."