HIGHLIGHTS
- who: Synthesizer Sound Matching and colleagues from the (UNIVERSITY) have published the research: Improving Semi-Supervised Differentiable Synthesizer Sound Matching for Practical Applications, in the Journal: (JOURNAL)
- what: The authors propose a novel training strategy where the network is fully trained using both parameter loss and spectral loss. The authors investigate in detail the relationship between synthesizer architecture and the difficulty of parameter estimation. To analyze the behavior of this differentiable synthesizer, the authors examine the values and analytic gradients of spectral loss on a synthetic benchmark in Section IV. From these results, the authors . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.