Discrete Coding and Masked Modeling for Text-to-Stereo Audio Generation
Abstract
Loading abstract...
Stereo Codec Reconstruction
Ten reconstruction cases comparing the reference stereo signal against Wavtokenizer-FT, Stable Audio Open, and StereoCodec.
Headphone recommended
Please use headphones for the best stereo listening experience.
| Ground Truth | Wavtokenizer-FT | Stable Audio Open | StereoCodec (Ours) |
|---|
Text-to-Stereo Generation
Eight text-conditioned stereo generation cases comparing SpatialSonic against StereoGen.
Headphone recommended
Stereo positioning and movement are easier to perceive with headphones.
| Caption | SpatialSonic | StereoGen (Ours) |
|---|