Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Bai, Liuyang, Lu, Weiyi, Guo, Li
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2512.21653
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911339082940416
author Bai, Liuyang
Lu, Weiyi
Guo, Li
author_facet Bai, Liuyang
Lu, Weiyi
Guo, Li
contents Speech codecs are traditionally optimized for waveform fidelity, allocating bits to preserve acoustic detail even when much of it can be inferred from linguistic structure. This leads to inefficient compression and suboptimal performance on downstream recognition tasks. We propose SemDAC, a semantic-aware neural audio codec that leverages semantic codebooks as effective priors for speech compression. In SemDAC, the first quantizer in a residual vector quantization (RVQ) stack is distilled from HuBERT features to produce semantic tokens that capture phonetic content, while subsequent quantizers model residual acoustics. A FiLM-conditioned decoder reconstructs audio conditioned on the semantic tokens, improving efficiency in the use of acoustic codebooks. Despite its simplicity, this design proves highly effective: SemDAC outperforms DAC across perceptual metrics and achieves lower WER when running Whisper on reconstructed speech, all while operating at substantially lower bitrates (e.g., 0.95 kbps vs. 2.5 kbps for DAC). These results demonstrate that semantic codebooks provide an effective inductive bias for neural speech compression, producing compact yet recognition-friendly representations.
format Preprint
id arxiv_https___arxiv_org_abs_2512_21653
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Semantic Codebooks as Effective Priors for Neural Speech Compression
Bai, Liuyang
Lu, Weiyi
Guo, Li
Sound
Computation and Language
Machine Learning
Speech codecs are traditionally optimized for waveform fidelity, allocating bits to preserve acoustic detail even when much of it can be inferred from linguistic structure. This leads to inefficient compression and suboptimal performance on downstream recognition tasks. We propose SemDAC, a semantic-aware neural audio codec that leverages semantic codebooks as effective priors for speech compression. In SemDAC, the first quantizer in a residual vector quantization (RVQ) stack is distilled from HuBERT features to produce semantic tokens that capture phonetic content, while subsequent quantizers model residual acoustics. A FiLM-conditioned decoder reconstructs audio conditioned on the semantic tokens, improving efficiency in the use of acoustic codebooks. Despite its simplicity, this design proves highly effective: SemDAC outperforms DAC across perceptual metrics and achieves lower WER when running Whisper on reconstructed speech, all while operating at substantially lower bitrates (e.g., 0.95 kbps vs. 2.5 kbps for DAC). These results demonstrate that semantic codebooks provide an effective inductive bias for neural speech compression, producing compact yet recognition-friendly representations.
title Semantic Codebooks as Effective Priors for Neural Speech Compression
topic Sound
Computation and Language
Machine Learning
url https://arxiv.org/abs/2512.21653