Abstract:

Effective speech representations for spoken language models must balance semantic relevance with acoustic fidelity for high- quality reconstruction. However, existing approaches struggle to achieve both simultaneously. To address this, we introduce Hierarchical Acoustic and Semantic Representation Disentanglement (HASRD, pronounced “hazard”), a framework that factorizes self-supervised learning representations into discrete semantic and acoustic tokens. HASRD assigns the semantic representation to the first codebook, while encoding acoustic residuals in subsequent codebooks. This preserves ASR performance while achieving high-quality reconstruction. Addi- tionally, we enhance HASRD’s encoder efficiency, improving ASR performance without compromising reconstruction quality. Compared to SpeechTokenizer, HASRD achieves a 44% relative WER improvement, superior reconstruction quality, and 2x lower bitrate, demonstrating its effectiveness in disentangling acoustic and semantic information.