WHITE BOX SEARCH OVER AUDIO SYNTHESIZER PARAMETERS

Synthesizer parameter inference searches for a set of patch connections and parameters to generate audio that best matches a given target sound. Such optimization tasks ben-efit from access to accurate gradients. However, typical audio synths incorporate components with discontinuities – such as sawtooth or square waveforms, or a categorical search over discrete parameters like a choice among such waveforms – that thwart conventional automatic differen-tiation (AD). AD libraries in frameworks like TensorFlow and PyTorch typically ignore discontinuities, providing in-correct gradients at such locations. Thus, SOTA parameter inference methods avoid differentiating the synth directly, and resort to workarounds such as genetic search or neural proxies. Instead, we adapt and extend recent computer graphics methods for differentiable rendering to directly differentiate the synth as a white box program, and thereby optimize its parameters using gradient descent. We evalu-ate our framework using a generic FM synth with ADSR, noise, and IIR filters, adapting its parameters to match a va-riety of target audio clips. Our method outperforms base-lines in both quantitative and qualitative evaluations.