Towards Lightweight Controllable Audio Synthesis with Conditional Implicit Neural Representations

Abstract

Table 1: Means of CDPAM, Multi STFT MSE and MSE for NSYNTH Diverse and NSYNTH Keyboard over all set items. Standard deviations are calculated over 3 training runs. For all metrics a lower score is better. Silence and noise (µ = 0, σ = 1) scores are reported for reference.