I tried not to tell that, but did you notice that convolver may be used as physical emulation synthesizer?
General model is identical to Sakura. The difference is that exciter, string and resonator are sampled which gives far more realistic simulation.
Let's say violin.
The exciter is bow. To sample it we may move it on the edge of hard material, i.e. desk and record this noise. Over all if this way of simulation wouldn't give results close to nature we may use a oscillator. For mallet instrument we may use digital dirac impulse lowpassed with filter according to mallet hardness. In general the impulse is plot of how the force changes in function of time. Therefore dirac for hit of mallet and noise for bow. Let's sample it and put into sampler.
Next is string. It's easy to sample it muting the resonator with clothes or removing it from resonator and stroking. Worth noticing is fact this is the only transposable part of an instrument and it may be simulated by oscillator too. If not, it is transposable without formant in resample-like manner. Sampler as first in chain would simulate it (with excitation in convolver as second in chain instead) from one sample if dirac used as exciter or convolver when other excitation happens. That's why physical simulation of instrument would take only few MB. So we sample string and put into convolver.
Next is box of a resonator. To sample it you may mute the strings and hit it with mallet. Resonator gives formant to instrument and it is main cause why we have to sample each note separately in case we want to get realistic sound of an instrument. In our case when the string is sampled with one sample and resampled we may sample resonator once to save many MB comparing to method when string with different pitches and resonator has to be sampled separately. So we put sample of resonator hit into next convolver in chain.
Next is environment, let's say different rooms with impulse responses recorded et.c. Easy to take it from convolver library or reverb or so.
Next may be head impulse responses for binaural effect of simulation. That's another convolver in chain.
Summarising we take everything into a insert channel each thing response in separate convolver and feed it with sample of excitation. It should sound very realistic. As a combination we may load up string response to sampler (if we want to play different notes from one sample) and exciter response to convolver as order of devices doesn't matter when convolving.
And it would do.
Exchange FL Studio tips & tricks
6 posts • Page 1 of 1
[You can only see part of this thread as you are not logged in to the forums]