Samuele Cornell

LACE (Layer-Adaptive Codec Encoding), a dynamic frame rate codec that applies an independent compression step at each quantization layer, enabling layer-specific segmentation boundaries, is proposed.

Downstream performance hinges on SSL pre-training language, not the NAC training language, enabling efficient cross-language model reuse.

Training speech separation models on real-world noisy data doesn't have to mean accepting noisy outputs: this method cuts residual noise in half.

Achieve single-pass alignment of multi-talker speech – a feat previously impossible – by modeling overlaps as shuffles.