Sujet Commentaires sur le test : Traitement de choc

I just want to share some tips concerning the wavetable-ization of acoustic instrumental and vocal sources, as the results are often poor when importing samples into Serum, depending on the overtone structure and the dynamics of a given sample.

Fist one should prepare the samples to import in a way, that all non-hamonic content is removes, e.g. in a sustained flute sample, remove all the breathing, this works best with a spectral filter like e.g. RX3 - e.g. in a trombone sample, remove the air hiss and the very high overtones, e.g. in a electric guitar power chord, remove as much of the amp noise as possible and so on. Then level the harmonics, subdue strong harmonics and enhance the root note.

But even with those steps, some instruments and vocal samples just won't resynthesize in any useful way unless you take the time to manually delete the garbled waveforms from the resulting wavetable, only keeping the substantial ones and then using the various morphing functions in Serum to generate a fluent wavetable.

I just spent about 40 minutes doing that with some male overtone singing, a simple transition from u -> i and no matter which import method I tried, the results were very garbled and crackly. I then listened to all 256 waveforms, kept about 15 of them and then morphed spectrally between them - the result was stunningly beautiful!

So, don't give up too early when an import doesn't yield instantly usable results, investing time and patience can bring out some very good and organic waveforms.

Everything metallic without a distinct pitch and a lot of odd harmonics is very hard to resynthesize in Serum though, but these sources can yield good results when using the FFT import or for generating single cycles which can then be combined in a wavetable or used as standalone.