Idea for realistic tremolo renders
Hello, this is my first post on the forums, and I hope I’m not posting in the wrong section. But there is this idea I had since a long time on how to better render tremolo and consecutive notes, and I’ve finally tested it myself.
The main issue with tremolo is that it plays the same sample in rapid succession, giving a "machine-gun" effect. Some soundfonts solve this problem by having multiple samples for each note, but here, it could be applied to any soundfont.
So, my idea is to apply a random amount of EQ to the samples each time one is played. In my demo, i used a mandolin. first you hear how it sounds on Musescore, and a bit later, I created my own tremolo from only one sample by adding to some of the notes an EQ that boosts the bass and lowers treble, and on others another that boosts the treble and lowers bass. Some are just the original sample. I think the results are convincing: and even when played for a sustained amount of time, it sounds somewhat realistic.
Ideally, the amount of EQ would be different and random each time, so no pattern forms.
But while this effect works very well with the mandolin, it works kind of bad on the staccato violin notes (last part of the demo sound file). maybe the effect is too strong, or maybe the EQ curve I chose isn’t ideal (but this was just for testing) and we could imagine a setting that would allow us to apply different strengths of "variable EQ round-robin" to different instruments.
If coupled with standard humanizing features, like velocity variation, and time variation, I think this would greatly improve the rendering of not only tremolos, but also strumming patterns, percussion parts like snare drum rhythms, and consecutive staccato notes. As tremolo is such an important technique on many instruments, I think it is important to render it well.
Thank you if you read all of this and I wish you good luck on Musescore 4 development!