more info required
was anything written down? any notation?
how were the mock-ups created?
at this current time the “Unmix Orchestra” does not exist; percussion Unmixing has greatly improved from SL11 to SL12.
Realize that Unmix Song separates into 7 layers:
- Vocals
- Drums
- Bass
- Guitar
- Piano
- Sax & Brass
- Other
I’d say, at this time in the unmixing dev, you are better off starting over. It’s a bitter pill, I know. SL could be used to display note values instead of freq to give a roadmap for re-writing the parts. The Spectrogram could give some useful note and timing info (although waveform is better at representing time), but I wouldn’t even try to unmix a string section at this time. The time spent waiting for the unmix modules to whittle thru and then y’all listen back, looking at the “other” layer and think “wha?…Why?” and then start trying to manually move each sound to it’s specific instrument? My goodness, it would take forever at this stage. Very likely faster to re-compose.
Unmixing the percussion could give some useful returns…or could have y’all scratching your heads. The source is very important in my testing. I have live and multi-tracked performances of three and four piece “rock” and “jazz” type stereo recordings of original music. One group is mainly cello, elec guitar and a Roland OctaPad. SL just doesn’t cope well with separating strings. More elaborate orchestrations just add to the already difficult task. SL will send legato and “singing” timbres to the Vocal layer; effects to the other layer, as a short example. My textural electric guitar parts often end up spread across guitar, vocal and other layers.
So, it’s probably best to try short selections of a couple of the mock-ups and investigate what you get with what SL does with the sound sets. SL is not good at separating synth sounds. I haven’t tried much sample based music either, so I can’t really speak to that.
So, if y’all can get SL successfully separating out percussion, that would be a win. Separating drums from “melodic” instruments and separating individual drums in SL12 is better than SL11 IME, yet, again, I have not found SL very good at non-rock band instruments.
For example, I have a live performance with vocal, elec guitar, electric bass (MM Stingray) and 5pc drum kit; the FOH recording is way out of balance (drums far too loud) with a rather poor vocal. Well SL is pretty good at removing vocals (not 100%, but very good) and one particular song is very STP sounding. Well the bass entirely goes to the drums layer as the bass timing is married to the kick. Separating out the freq area where the bass and BD are would take ages manually or with “select similar”. That bass could be played in lot quicker than any SL separation techniques.
Similarly, I have a live folkish rock band with 5 vocals, at least 3 string players (guitars, banjo, mando, pedal steel), sax, electric piano (Ensoniq KS-32) and real accordion; recording quality similar to that STP sounding song. Well, the bass and drums separated very well in SL11 (I didn’t try SL12 yet); in this case, the balance is a bit better in the recording (drums not over loud) and the bass is a Fender P-Bass. Sax separates well in some areas, and not well in others: maybe some lead lines could be carved out…but by then, you have played them in. Elec Piano and synth are always troublesome. I haven’t tried any real piano or the accordion.
I’d say if you aren’t already an advanced manual editor using SL, that you’ll waste more time learning SL manual separation skills and how to read the spectrogram than if your friend just starts back at square one using those mock-ups as a guide. You’ll just spend so much time experimenting with SL, which would have been time better spent tracking.
