Posted
The user sings or plays a monophonic melody with guitar, flute, etc. into the mircophone, or via Line In.
Synfire records the audio, recognizes the note values (rhythm) and the pitches - taking into account the current harmony, which reduces the number of possible pitches - and then creates the figure on the fly !
Even if the pitch detection would not be too accurate, you would take advantage of the recognized rhythm and note durations, as this is the most tedious part of melody noation/figure creation.
Yes, this is not a trivial one, but synfire constists of many non-trivial features ;-)
Wed, 2015-01-21 - 19:53 Permalink
My perspective on composing has always been that of a singer, so yes, I agree this would be a great feature. It's not easy to so, though. Those algorithms have been around for decades and still have their issues.
Melodyne works great and has been integrated with many DAWs already.
Also check out Imitone. It's cheap and according to the discussion around it, it seems to work fine.
Wed, 2015-01-21 - 20:42 Permalink
Yes, the algorithms are not perfect and probably never will be.
But synfire would have one big advantage over these programs you mentionend:
It would just need to recognize the rough figure /movement, not the actual pitch. And the latter is the real challenge: Decide between 2 semitones. But with synfire it would just be a great fun to sing along to a harmony and get interesting results, just like painting the figure with the mouse.
