dGSLM samples

In this page we show the dGSLM outputs used in the paper

Please note that the first 5 seconds was prompted with the reference audio, and the rest of the audio was generated by the dGSLM model.