Physically modelled strings first attempt

SPAN Cello c3.vital (360.3 KB)
SPAN Double Bass c2.vital (360.3 KB)
SPAN Violin 1 c6.vital (435.3 KB)

Here are my first attempts at physically modelling string instruments.

I analysed real samples, edited wavetables to match the spectrum, filtered noise and then balanced the two. However, they only sound reasonable at the note I created them at.

Has anyone had much luck with this method?


I have tried. I think going all the way shaping the harmonic content of the oscillator AND then using both a comb and an LP is kind of overkill. Had made a video on the topic some time ago:

Ah… and also the following one, for kind of obvious reasons :smiley:

1 Like

the formant of the sample could probably be modified along the note scale if you wanted it to sound a little better as you play across the range. for example, when played low, increase the formant effect at the oscillator’s own selectable filter and if it’s a super high note, lower the formant. this is just a guess or maybe another thing to try. otherwise it might be worth it to look at what the main differences are between waveforms played at different notes to see whats happening to the waveforms. maybe you can make a wavetable and morph that according to the note being played (if you use the right wavetable modes etc). The tutorial by platipo is good too as an alternative. i’ve also seen comb filters being used to do violin type sounds.

Then again there’s a difference in tone at different registers on the same string, and also between the strings. Violinists use those all the time to shape the tone.