Unlike [17], we do not use the wave-based discrimina- tor [18]

LEARNING The model is trained with a combination of adversarial, reconstruction losses similar to [1], as illustrated in Figure 3 · 2048

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

browse 1 citing papers

citation-role summary

other 1

unclear 1

cs.SD · 2025-08-07 · conditional · novelty 6.0

SpectroStream, a 2D time-frequency neural codec, reconstructs 48 kHz stereo music at 4-16 kbps with better ViSQOL and subjective quality than DAC.

Showing 1 of 1 citing paper.

SpectroStream: A Versatile Neural Codec for General Audio cs.SD · 2025-08-07 · conditional · none · ref 3
SpectroStream, a 2D time-frequency neural codec, reconstructs 48 kHz stereo music at 4-16 kbps with better ViSQOL and subjective quality than DAC.