Conditional Sound Generation Using Neural Discrete Time-Frequency Representation Learning | IEEE Conference Publication | IEEE Xplore