Hey,
I have a song that I accidentally backed up the wrong folder for, losing the project file that contained all of the MIDI data for a string arrangement for that song.
I have a stereo mixdown of the track, but I really wanted to go back and refine the mix as I wasn’t 100% happy with it. I really loved that string arrangement though!
Using SL10 pro, I was able to extract the strings from the unmixed stereo render, and use some of the tools on SL to further isolate the string arrangement. It sounded better and cleaner than the other solutions that I tried.
What I need now, is a reliable, intelligent process that can convert the extracted audio to MIDI data, so that I can recreate the string arrangement with the same sample library.
There are some other solutions for translating audio to MIDI, but none of them are very refined or intelligent processes, and they require a LOT of editing of the note data to make it useable.
It would be awesome if SL was capable of performing this task on unmixed stems.
I think this is something we are all waiting for …
Two other products you could try (trial) are Melodyne and RipX, however in my experience the results are barely usable … depending on the source material (as always), it can be close enough to save some time, or completely random and an utter waste of effort.
For the purposes of transcription to MIDI, my experience has been that it’s often less effort to do it old skool by ear than it is to learn the idiosyncrasies of the current crop of software. It’s a different proposition to sound separation, which all of these are now quite good at.
Waiting anxiously for the iminent release of SpectraLayers 11 …
If you also like to study and decompose a song you can try DECODA. We have used it to deconstruct folk songs to midi for documentary archival purposes…
