During his internship at Dolby, Enric run an exhaustive evaluation of various loss functions for music source separation. After evaluating those losses objectively and subjectively, we recommend training with the following spectrogram-based losses: L2freq, SISDRfreq, LOGL2freq or LOGL1freq with, potentially, phase- sensitive objectives and adversarial regularizers.
We investigated various upsampling layers to consolidate the ideas we introduced in our previous paper. We benchmarked a large set of upsampling layers for music source separation: different transposed and subpixel convolution setups, different interpolation upsamplers (including two novel layers based on stretch and sinc interpolation), and different wavelet-based upsamplers (including a novel learnable wavelet layer).
PLAAE (packet loss adversarial auto-encoder) is our proposal for packet loss concealment in a non-autoregressive fashion. Our goal is to reconstruct missing speech packets until a new (real) packet is received in a video-call. Our end-to-end non-autoregressive adversarial auto-encoder specially shines at long-term predictions, beyond 60ms. The paper has been accepted for presentation at WASPAA 2021! Check out our arXiv pre-print.
Actually, what I really need is less papers with “all you need” in the title – and to share a (non-virtual) beer with you folks!! Here some of the papers I enjoyed, together with the papers we presented. You’ll see that I don’t include classification/tagging papers, I guess I need a break from my PhD topic 🙂 Enjoy!