Humanize audible effect, please!
Hi, Gang!!!
I think, one of the things MuseScore has forgot is the capability to add a random "humanize" audible effect about the exact "click" time of each played note.
I'm talking about this:
1) It is absolutely impossible to human being players to play (sound) in the mathematical exact starting moment which is indicated in the score.
2) That fact is the main reason to the typical "human" sound, against the "computer" sound (the machine is mathematically exact, any time you use it).
How to avoid it?
Simple! (To say): Adding a minimal (milliseconds) random time at the beginning of each played note.
And, of course, that audible effect should be an automatic process. If the user would want to use it: put it "ON"; if not, put it "OFF". Simple!
Could this "effect" be included in the new 4.x version, please? ???
Just a crazy idea, from an old man.
Blessings and Greetings from Chile!!!
Juan
Comments
Randomizing is not the same as humanizing and the errors made by humans are less random than you'd think.
There was fairly recently a more elaborate topic on the forums including links to some nice studies about what "humanizing" would entail.
Indeed, humans are far from random! But adding some humanization - using AI to actually mimic the patterns predictably applied by humans - is a huge part of the effort going into MuseScore 4. Specifically, the project known for now as "MuseSampler", although I hear that name may change before release.
In reply to Indeed, humans are far from… by Marc Sabatella
AI playback in MS4? This sounds very encouraging.
In reply to AI playback in MS4? This… by yonah_ag
See the various progress reports in the Announcements forum, and pay particular attention to the discussions of playback, NotePerformer, and MuseSampler. It's still too early to say exactly what MuseSampler will entail, but it does seem clear it is meant to be an in-house alternative to NotePerformer.
See related discussion here:
https://musescore.org/en/node/298260
I have tested random OnTime for notes and it doesn't reduce the robotic effect of computer playback. If the duration is really minimal then it is not heard and the playback does not improve. If it is long enough to be heard then the playback sounds erratic and sloppy.
I think that our brains cope with the lack of mathematical precision in human playing and basically ignore trivial deviations from perfect time.
But see the discussion in the link above for other opinions and ideas for improving the musicality of computer playback.
In reply to See related discussion here:… by yonah_ag
Well, this is not only about the starting point variations.
Any musical note, whatever the instrument, has the famous ADSR (Attack, Decay, Sustain and Release) curve, and ALL of those "parameters" are different and unique to each human player.
Can a machine reproduce this phenomenon? ???
In reply to Well, this is not only about… by jotape1960
AI programming would be needed but such programs are written by humans and are trying to emulate human playing so, fortunately, there's no getting away from us pesky humans!
The best that my plugout can hope to achieve is a reduction in robotic playback. It is not capable of any humanisiation – but the info being released about MS4's playback sounds like we're in for a huge leap forward so my plugout will more than likely be redundant soon.
In reply to Well, this is not only about… by jotape1960
The mechanical-effect occurs most often on instruments that, after playing, cannot contribute to the control of the sound (self-damping). Piano, Guitar, Percussion etc.
ADSR (or AHDSR) This is about tailoring the sound, not about "humanize".
What you are talking about is about instrument playing-style and nuances (volume-effects). Which are about instruments where you can control (hold) the sound in sustain-time. eg: Strings, Woodwinds, Brass, etc.