CS Paper Radar

Daily listing

2026-11-17 · cs.SD

4 papers first submitted this day (UTC) in cs.SD — Sound. arXiv announces a day's submissions the next evening, so the newest day fills up over ~24 h.

Score = weighted mean ofCitations 20%Upvotes 10%Code stars 10%Venue 15%Code/data 10%Novelty 15%Momentum 15%Substance 5%how it works

Language, Vision & Media 4

High-Fidelity Binarized Neural Vocoder

cs.SD ·2026-11-17 ·Andong Li, Tong Lei, Xiaodong Li, Chengshi Zheng
publishedcode/data
Abstract · Accepted by IEEE Transactions on Audio, Speech and Language Processing 2026, 15 pages, 10 figures, 16 tables

Neural vocoders have achieved remarkable progress in high-fidelity waveform generation in recent years, but most existing methods still rely on full-precision operators, leading to non-negligible storage and computational costs during inference. Binary neural networks (BNNs) provide an alternative direction by quantizing weights and activations to 1 bit. However, directly applying binarization to neural vocoders is rather challenging as it inherently incurs severe information loss. To this end, we propose BiVoC, a BNN-oriented time-frequency neural vocoder for high-fidelity waveform generation. Building on our previous full-precision RNDVoC framework, BiVoC adapts the inverse-problem formulation, range-null decomposition, and fineto-coarse non-uniform subband modeling to the 1-bit setting. Specifically, we develop a simple binarized convolutional unit to provide stable time-frequency feature modeling under binary representations. Furthermore, we analyze the quality-complexity sensitivity of different modules and propose a mixed-precision design to retain quality-sensitive components in full precision while binarizing computation-dominant parts. Extensive experiments are conducted on LibriTTS and multiple out-of-distribution benchmarks. Both objective and subjective results demonstrate that BiVoC substantially outperforms BNN variants obtained by directly binarizing existing vocoders, while achieving competitive performance against several mainstream full-precision vocoder baselines under much lower theoretical computational cost. These results demonstrate the promising potential of BNNs in the audio generation field. Codes and pre-trained weights are available at https://github.com/Andong-Li-speech/BiVoC.

0.77

Insight into Parametric First-Order Ambisonics Upscaling in Time and Frequency Domains

eess.AS cs.SD ·2026-11-17 ·Oded Levi, Boaz Rafaely
Abstract · 10 figures. Listening-test data: https://doi.org/10.5281/zenodo.24257365

First-Order Ambisonics (FOA) offers a practical representation for spatial audio, but its low-order reproduction introduces distinct perceptual limitations during binaural rendering. While recent methods synthesize Higher-Order Ambisonics (HOA) from FOA to overcome these constraints, the importance and potential of early reflections in the upscaling process have not yet been comprehensively explored. To bridge this knowledge gap and understand the potential of incorporating early reflections in FOA upscaling, this work develops an experimental framework that selectively upscales these reflections. By utilizing an energy-based criterion, the proposed framework identifies and selectively renders the strongest early reflections with high-order resolution. The selection mechanism is evaluated across three domains: long-term temporal, short-time temporal, and the joint time-frequency (TF) domain. Validation through objective metrics, including the Scale-Invariant Signal-to-Distortion Ratio (SI-SDR), the clarity index ($C_{50}$), and Interaural Cross-Correlation (IACC), alongside listening tests, demonstrates that the TF-domain provides superior performance. By exploiting the sparsity of speech, the TF approach maintains robust performance in complex multi-talker scenarios, providing insights into the benefits of early reflections for the design of future parametric and data-driven FOA upscaling methods.

0.34

Beyond Mimicry: Constrained Exploration with GRPO for Joint Multi-Talker ASR and Diarization under Unknown Speaker Counts

cs.SD ·2026-11-17 ·Yunrui Cai, Dingdong Wang, Lingwei Meng, Xixin Wu, Zhiyong Wu +1 more
published
Abstract · 5 pages, Interspeech 2026

Joint multi-talker ASR with speaker attribution and timestamps is particularly challenging when the speaker count is unknown and the output must follow a strict structure. We observe that Supervised Fine-Tuning (SFT)-only alignment of SpeechLLMs is brittle under dense overlaps, often leading to burst hallucinations and malformed speaker/timestamp tags. We propose a unified two-stage generative framework: (i) CoT-augmented SFT that infers the global speaker count before transcription, and (ii) GRPO-based constrained exploration guided by a Multi-dimensional Constraint-Aware Reward (MCAR) that directly optimizes permutation-invariant accuracy while enforcing counting, temporal, and structural constraints. On Libri3Mix and a Dynamic-Mix (2+3) set (speaker count unknown at test time), our model achieves 14.52%/9.24% cpWER and 1.95%/1.12% WDER, respectively, with GRPO yielding a 35%/54% relative cpWER reduction over SFT+CoT.

0.31

Back to Mel: Learning to Bridge Fidelity and Perception in Generative Speech Enhancement

cs.SD ·2026-11-17 ·Andong Li, Kun Zhou, Xiaodong Li, Chengshi Zheng
code/data
Abstract · Submitted to IEEE TPAMI, 18 pages, 15 figures, 16 tables

Recent generative techniques have facilitated the rapid proliferation of speech enhancement approaches built upon learnable compact latent representations and large generative backbones. Nevertheless, it remains inherently challenging to balance between acoustic fidelity, perceptual quality, and processing efficiency. In this paper, we revisit the Mel spectrum as a structured acoustic representation and propose BAME, a flow-matching-based framework that exploits the complementary priors for generative speech enhancement task. First, leveraging the explicit hierarchical frequency distribution of Mel spectrum, we elaborately devise a subband-aware Transformer (SaT) for efficient generative modeling. Using only 2.9% of the parameters, the proposed SaT achieves superior enhancement performance compared with a 331 M DiT backbone. Second, motivated by range-null decomposition theory of Mel filtering in the magnitude-spectral domain, we introduce a phase-aware neural vocoder that incorporates degraded-phase prior, together with a quality-guided dual-phase strategy to adaptively mitigate the effects of unreliable phase information. Third, we devise a simple yet effective one-step fidelity-perception alignment strategy that fine-tunes the flow model with complementary constraints from fidelity and perception domains. Extensive experiments are conducted on DNS-Challenge, EARS, GSR, and URGENT benchmarks, and both objective and subjective results demonstrate promising performance on various intrusive and non-intrusive evaluation metrics. These results further advocate the potential of exploiting structured acoustic representations for reliable and efficient generative speech enhancement. Audio samples and codes are available at https://github.com/AndongLi-speech/BAME.

0.25