arXiv_SD
-
Generalised Transcoding Framework for Arbitrary Spatial Audio Capture and Playback Formats
Abstract
This article introduces a unified framework for the parametric analysis and reproduction of spatial sound scenes captured either as Ambisonic signals or as raw microphone array signals. The proposed method estimates time-frequency-dependent spatial metadata that characterises a variable number of primary source components and an ambience component with its own angular power distribution, whose parameters fit the observed spatial covariances of the captured signals. This metadata is used to construct spatial covariances of the target playback formats, which are then used to derive optimal mixing matrices for transcoding the scene for playback over the target reproduction system. The method additionally handles independent rotations of both capture and playback setups. Real-time implementations of the method and other existing state-of-the-art parametric renderers are compared in a listening test using simulated scenes from Ambisonic, spherical, and head-worn arrays. The results highlight perceptual benefits of the proposed framework across a diverse range of content and receiver configurations, particularly for lower-order and geometrically constrained microphone arrays.
Abstract (translated)
URL
https://arxiv.org/abs/2606.18480
PDF
-
Next-Turn: Duration-Aware Streaming Endpoint Detection via Time-to-Next-Speech-Onset Prediction
2026-06-16 15:58:36Tristan Tsoi, Jiajun Deng, Yingke Zhu, Huu Quyen Dang, Tianxiang Cao, Nikita Kuzmin, Tao Zhong, Simon LuiAbstract
Endpoint detection (EPD) is essential for natural turn-taking in streaming speech systems. However, reliably determining the endpoint of an utterance is challenging because speakers often pause mid-utterance due to hesitations and disfluencies. Semantic EPD has emerged as a promising direction to address this issue but is hindered by ambiguous supervision and strict streaming constraints. We propose Next-Turn that uses the time-to-next-speech-onset as the training objective, where targets are derived directly from speech timestamps and require no additional annotation. Experiments show that the proposed method outperforms conventional acoustic and recent semantic EPD baselines, achieving a 25.9% absolute improvement in endpoint accuracy within 320 ms over the strongest baseline. In addition, joint training with the duration-aware objective complements standard binary EPD, with gains that increase monotonically with increasing pauses.
Abstract (translated)
URL
https://arxiv.org/abs/2606.18094
PDF
-
Reliable Neural-Codec Text-to-Speech by ASR Self-Verification and Distillation: Near-Zero Catastrophic Failures Across Models and Codecs
2026-06-16 15:41:44Ali Asaria, Tony Salomone, Deep GandhiAbstract
Open autoregressive neural-codec text-to-speech (TTS) models sound excellent on typical inputs yet suffer stochastic catastrophic failures: on a meaningful fraction of utterances they emit silence, terminate early, or collapse into repetitive or hallucinated content. We show this failure mode is cheap to remove. Under a single format-robust metric (a catastrophic-failure rate via an ASR round-trip), best-of-N ASR self-verification drives failures to near-zero: no observed failures remain by N=2 on a standard corpus (LibriSpeech) and by N=4 on a hard prompt set. This is not an artifact of one model: the reduction replicates across four open codec-TTS systems and three neural codecs (XCodec2, SNAC, Mimi), reaching the near-zero floor by N=2 on three of the four. We then make the fix free at inference time by distilling the self-verified behaviour into the model, which recovers much of the robustness in single-shot decoding, closing ~52-58% of the failure mass on hard inputs at no test-time cost. The distillation gain concentrates where it is needed (hard inputs); on already-reliable prose there is no headroom and no detectable change. A controlled comparison adds a clean negative: offline direct preference optimization (DPO/IPO) does not beat plain supervised distillation, and an online iterative variant is promising but not statistically separable at our evaluation size. We report honestly the one model that resists (a larger Llasa where scale did not obviously help) and a rare-word capability ceiling that no self-distillation method overcomes
Abstract (translated)
URL
https://arxiv.org/abs/2606.18323
PDF
-
DeSRPA: Decoupled Speech Role-Playing Agent via Inference-Time Intervention
2026-06-16 08:30:47Wenqiu Tang, Zhen Wan, Takahiro Komamizu, Ichiro IdeAbstract
While Large Language Models (LLMs) have revolutionized text-based role-playing, creating immersive Speech Role-Playing Agents (SRPAs) requires a seamless bridge between cognitive reasoning and paralinguistic nuances. Current SRPAs primarily rely on end-to-end (E2E) fine-tuning. However, this paradigm suffers from poor generalization to unseen characters due to its reliance on role-specific data, while imposing a "modality alignment tax" that degrades intrinsic LLM reasoning capabilities. We propose DeSRPA, an agentic framework for character role play via inference-time intervention on frozen backbones. DeSRPA employs a dual-level control vector mechanism, Internal Cognitive Steering and External Expressive Rendering, to synchronize "mind" and "voice". Experiments on SpeechRole and OmniCharacter benchmarks demonstrate that DeSRPA significantly outperforms E2E baselines in personality and emotional consistency. It achieves high speech naturalness, narrowing the gap with proprietary models like GPT-4o Audio, while remaining a scalable and training-free paradigm.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17669
PDF
-
A Closer Look at Failure Modes in Temporal Understanding of Large Audio-Language Models
2026-06-16 01:57:56Apoorva Kulkarni, Kaousheik Jayakumar, Sreyan Ghosh, Sarah Wiegreffe, Dinesh Manocha, Ramani DuraiswamiAbstract
Large Audio Language Models (LALMs) achieve strong performance on a variety of audio understanding tasks but continue to struggle with temporal reasoning, a fundamental capability central to human auditory perception. Understanding the causes of these failures remains challenging as existing benchmarks report performance gaps without probing underlying mechanisms. To address this, we introduce a benchmark with 1,657 questions across three foundational tasks designed specifically for mechanistic analysis. Examining model outputs across varying input settings (behavioral analysis) reveals that models often under-utilize audio when textual cues are available. We also provide the first causal mechanistic analysis of temporal reasoning failures in LALMs. Comparing attention upweighting against scaling, we find that redistributing attention across audio tokens is more effective than increasing audio attention. Targeting task-relevant tokens yields further gains. These findings suggest that modality imbalance alone cannot explain failures. Attention scaling at bottleneck layers improves accuracy from 55.9% to 59.1% without fine-tuning, demonstrating a promising direction for future work.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17417
PDF
-
ELSA: Acoustic Event-Level Semantic Alignment for Fine-Grained Reference-Free Text-to-Audio Evaluation
2026-06-16 01:38:38Shuntaro Suzuki, Kento Tokura, Daichi Yashima, Kanon Amemiya, Komei Sugiura, Shinnosuke TakamichiAbstract
Text-to-audio (TTA) generation, synthesizing audio from natural language, has been widely studied for its ability to capture precise user intent. To effectively advance TTA models, it is essential to reliably evaluate generated audio without relying on costly human subjective ratings, motivating the development of automatic evaluation metrics that correlate well with human judgments. While recent CLAP-based metrics provide practical reference-free solutions, their coarse-grained text-audio similarity matching often correlates poorly with human ratings. To address this, we propose ELSA, a reference-free evaluation metric for fine-grained text-audio alignment. ELSA decomposes generated audio guided by distinct acoustic events derived from the text query and assesses event-level alignment. Experiments across four TTA benchmarks show that ELSA reveals a higher correlation with human subjective ratings than prior metrics, highlighting its effectiveness for reliable TTA evaluation.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17404
PDF
-
Turning music identification into a neural forward pass
2026-06-15 21:12:44Muhammad Taimoor Haseeb, Ahmad Hammoudeh, Gus XiaAbstract
Search, a foundational operation in computer science, maps a query to a matching item in a collection. It is typically implemented as a System-2 like, rule-based pipeline in which a key is computed, an index is probed, and candidates are verified. By contrast, human recognition resembles a System-1 like, associative model of identity recovery, in which even partial cues can trigger a recall without explicitly enumerating, ranking, or even accessing discrete candidates. Here, we show that music sound identification, a difficult search problem, can be performed in a single neural feed-forward pass by a generative transformer. Trained on an audio dataset, the model predicts the corresponding track identifier from a short audio excerpt. This approach surpasses state-of-the-art acoustic fingerprinting, with the largest gains for short audio segments (1 second), demonstrating the method is not only viable but advantageous. Moreover, it reduces external storage to 0.33% of the baseline footprint and improves inference latency by 2.3x (p95). Furthermore, the model can reject queries for unseen tracks, supporting open-set operation while reducing misattribution risk. Using music track identification as an example, this work reframes search, bringing it closer in spirit to human associative recognition and away from algorithmic database lookup.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17301
PDF
-
Intelligibility of Speech in Noise: Investigating Contribution of Magnitude and Phase Spectra
2026-06-15 20:07:37Bhanu Teja Nellore, Sudarsana Reddy Kadiri, Rohit Kumar, Karan Nathwani, Suryakanth V GangashettyAbstract
It is well known that intelligibility of speech reduces in the presence of ambient noise. However, studies show that all sounds are not affected uniformly (or equally) and that vowels are more robust to noise than consonants. In this study, intelligibility of various consonants is assessed and analyzed in stationary white noise and non-stationary babble noise conditions. Specifically, this study investigates the individual contribution of magnitude and phase spectra of a given speech signal on human speech recognition of consonants in noisy conditions. In this regard, three experiments are carried out. In experiment 1, clean signal, signal reconstructed with only magnitude spectrum information (magnitude only signal) and signal reconstructed with only phase spectrum information (phase only signal) are assessed for intelligibility. In experiment 2, noise is added to clean speech. From noisy speech, phase only signal and magnitude only signal are reconstructed and intelligibility tests are performed for all these three signals. In experiment 3, noise is added directly to the magnitude only and phase only signals reconstructed from clean speech and their intelligibility is assessed. Results of these experiments show that magnitude spectrum contributes more to intelligibility in clean condition than phase spectrum, while information from phase spectrum is more robust in noisy conditions. It is also observed that, among consonants, nasals are more susceptible to noise whereas fricatives and approximants were observed to be comparatively more robust.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17259
PDF
-
Transductive Zero-Shot Audio Classification with Audio-Language Models
2026-06-15 18:01:08Jingwen Zhou, Mingzhe WangAbstract
Contrastive language-audio pretraining (CLAP) enables zero-shot audio classification, but standard inference classifies each clip in isolation and ignores the structure of the unlabeled test set. We present the first systematic study of TransCLIP-style transductive inference for CLAP: a text-anchored spherical Gaussian-mixture EM that refines zero-shot posteriors using the audio-embedding statistics of the test batch, with no labels, no gradients, and negligible compute (about 15 ms on one CPU core for 2,000 clips). Across ESC-50, UrbanSound8K, and VocalSound, this consistently improves top-1 accuracy by +4.6 to +9.2 points over the zero-shot baseline (e.g., 89.1 -> 94.8% on ESC-50, 73.8 -> 81.8% on UrbanSound8K). We further show that the gain (i) is governed by a simple operating boundary -- roughly 2.5 test samples per class per batch are required, with diminishing returns beyond ~5; (ii) is complementary to entropy-guided prompt weighting, with the combination reaching 96.2% on ESC-50; and (iii) attenuates but remains positive under long-tailed batches (+4.9 -> +3.1 points at a 20:1 imbalance), which we report as an explicit limitation. We also document a negative result: on TUT Urban Acoustic Scenes 2018, where zero-shot CLAP is near chance, transduction has no signal to amplify.
Abstract (translated)
URL
https://arxiv.org/abs/2606.17160
PDF
-
CraBERT: Efficient Phoneme Encoder Pre-Training via Cascade Fusion of Subword Representations for Text-to-Speech
2026-06-15 13:04:25Dong Yang, Yuki Saito, Wataru Nakata, Hiroshi SaruwatariAbstract
This paper introduces CraBERT, a pre-trained phoneme encoder (PPEnc) designed for efficient pre-training in text-to-speech (TTS). CraBERT employs a cascade-fusion architecture and a subword-phoneme alignment algorithm to integrate representations from a pre-trained subword-level BERT into a phoneme-level BERT. This design provides prior word- and sentence-level information, reducing the amount of pre-training required by the phoneme encoder. Subjective listening evaluations show that CraBERT achieves MOS values comparable to existing PPEncs after approximately one epoch of pre-training, whereas the baselines in our comparison are pre-trained for approximately ten epochs. These results demonstrate that CraBERT can efficiently learn representations suitable for improving the perceived naturalness and prosody of synthesized speech.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16668
PDF
-
Beyond Artifacts: Towards Generalizable Synthetic Song Detection via Music-Intrinsic Features
2026-06-15 12:02:53Yan Han, Zhibin Wen, Yuan Wang, Shuangrun Shao, Xiaobing Li, Yang Xu, Wei LiAbstract
The rapid advancement of AI music generators highlights the urgent need for reliable Synthetic Song Detection (SSD). Existing SSD methods often rely on low-level artifacts or fixed feature assumptions, struggling to capture generator-agnostic cues. To address this, we propose Sofia (Synthetic-song detection framework via music features), a flexible framework that models music-intrinsic attributes via feature-specific experts and an adaptive Mixture-of-Experts (MoE) module. By configuring Sofia with representative Vocal, Audio-effect, Global structure features, and their combinations, we present their individual and complementary contributions. To comprehensively evaluate our framework, we further construct MUSIC8K, a challenging benchmark featuring lastest emerging generators and realistic audio perturbations. Experiments show that Sofia learns generator-agnostic representations from music-intrinsic features, improving the F1 score by 18.5 points over the strongest baseline on MUSIC8K-O while maintaining strong robustness.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16612
PDF
-
Confidence Score Guided Incremental and Speaker Adaptive Pseudo-Labeling for Semi-Supervised Elderly Speech Recognition
2026-06-15 10:49:46Chengxi Deng, Xurong Xie, Shujie Hu, Jiajun Deng, Mengzhe Geng, Youjun Chen, Huimeng Wang, Haoning Xu, Guinan Li, Xunying LiuAbstract
This paper proposes a novel confidence score guided incremental and speaker adaptive pseudo-labeling approach for semi-supervised elderly speech recognition. It facilitates higher-quality pseudo-label selection and progressive refinement, while also mitigating speaker heterogeneity. A confidence estimation module is designed to rank the reliability of untranscribed data, enabling a curriculum learning trajectory that progressively folds in unlabeled data subsets from high to low confidence. Speaker-specific characteristics are captured through speaker adaptive training with learnable prompts. Experiments on the English DementiaBank Pitt and Cantonese JCCOCC MoCA elderly speech datasets suggest that the proposed method outperforms the semi-supervised baseline using no confidence scores guided incremental or speaker adaptive pseudo-labeling by statistically significant word error rate (WER) or character error rate (CER) reductions of 1.45% and 2.27% absolute (6.21% and 6.98% relative).
Abstract (translated)
URL
https://arxiv.org/abs/2606.16546
PDF
-
Decoding while Adapting: Zero-Shot Online Speaker Adaptation via Audio-Textual Prompts for Elderly Speech Recognition
2026-06-15 10:40:48Chengxi Deng, Xurong Xie, Shujie Hu, Mengzhe Geng, Tianzi Wang, Youjun Chen, Huimeng Wang, Haoning Xu, Jiajun Deng, Xunying LiuAbstract
This paper proposes a novel cross-utterance audio-textual prompts based speaker adaptation approach for elderly speech recognition. It enables zero-shot, real-time adaptation to unseen speakers. Speech and text embeddings are extracted from the current and a few preceding utterances, before being fused in a cross-modal manner to produce compact speaker prompts that are more consistent than i/x-vectors and ECAPA-TDNN features. Experiments on the English DementiaBank Pitt and Cantonese JCCOCC MoCA elderly speech datasets suggest that the proposed online adaptation outperforms the speaker-independent (SI) model by statistically significant word error rate (WER) or character error rate (CER) reductions of 0.61% and 1.22% absolute (2.99% and 4.48% relative). Real-time factor (RTF) speed-up ratios of up to 9.83 times are obtained over offline batch-mode adaptation.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16539
PDF
-
Semi-Supervised Speech Confidence Detection using Pseudo-Labelling and Whisper Embeddings
2026-06-15 10:06:50Adam Wynn, Jingyun Wang, Xiangyu TanAbstract
Understanding speaker confidence is crucial in educational settings, as it can enhance personalised feedback and improve learning outcomes. This study introduces a novel framework for detecting speaker confidence by integrating human-engineered features with embeddings from the Whisper encoder. To address data limitations, a pseudo-labelling technique is employed to expand the labelled dataset, allowing the model to learn from both human-annotated and model-generated labels. The framework combines traditional speech features including pitch, volume, rate of speech, and the presence of disfluencies and stress, with Whisper embeddings, and uses a co-attention mechanism to fuse these representations and achieve an overall accuracy of 75%. This study contributes to advancing speech analysis, enabling applications that support personalised learning and speaking skill development.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16505
PDF
-
Towards Robust Generative Speech Enhancement Using Vector Quantisation-Based Neural Audio Codec
Abstract
This work investigates modelling strategies in continuous and discrete latent spaces in the vector quantisation (VQ)-based neural audio codec (NAC) speech enhancement (SE), along with the role of VQ regularisation. We propose cNAC-SE and dNAC-SE frameworks that predict continuous representations and discrete tokens in latent space, respectively. Theoretical analysis and visualisations in latent space are performed to exhibit their inherent modelling mechanisms. Experimental results show that the fully fine-tuned cNAC-SE model consistently outperforms all dNAC-SE variants across diverse test conditions and achieves leading performance among established generative approaches in DNS-MOS metrics. Comparison with the discriminative counterpart shows that VQ enhances robustness through an intrinsic effect of clean-prior-constrained regularisation, independent of discrete token processing. This highlights the transferable value of VQ regularisation to other continuous modelling methods.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16464
PDF
-
Unified Audio Generation and Editing via Joint Condition Modeling and Progressive Training
2026-06-15 09:06:38Haocheng Dong, Yuheng Lu, Cheng Gong, Shansong Liu, Xiao-Lei Zhang, Xuelong LiAbstract
With the growing focus on audio in multimedia applications, numerous advanced works on audio generation have emerged. Existing studies typically treat text-to-audio (TTA) and other related audio generation tasks, such as instruction-based audio editing, as independent challenges, adopting task-specific architectures or modules. This absence of a unified modeling paradigm substantially increases the overhead and complexity of building a system for both audio generation and editing, while also leading to limited scalability. To address this issue, we introduce AudioWeave, a unified model for TTA and audio editing without additional task-specific components. Specifically, we propose a joint condition modeling approach with a factorized position embedding, enabling the diffusion transformer backbone to operate under heterogeneous inputs of TTA and audio editing. We further propose a progressive multistage training strategy to mitigate task competition and catastrophic forgetting caused by interference among multiple tasks. This in turn helps maintain the performance of each individual task and may even lead to improvements in certain aspects. Experimental results on TTA task and six audio editing tasks show that our unified model achieves competitive performance with task-specific models, laying a groundwork for further exploration of unified audio generation models.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16435
PDF
-
Joycent: Diffusion-based Accent TTS without Accented Phone Prediction
Abstract
Accent text-to-speech (TTS) aims to synthesize speech with target accents. Existing accent TTS systems typically rely on a two-stage pipeline that first converts standard phone sequences into accented phone sequences and then synthesizes accented speech. However, such approaches suffer from error accumulation and require paired standard-accented phone sequence data, which is often limited in practice. Moreover, text-based accented phone representations are insufficient to model acoustic accent characteristics such as prosody and rhythm. In this work, we propose Joycent, a diffusion-based accent TTS model that synthesizes accented speech directly from standard phone sequences and speech references without accented phone prediction. Joycent integrates accent and speaker representations through conditional layer normalization (CLN) in the text encoder. We introduce WhisAID, a Mandarin accent identification model trained on accented Mandarin speech to extract accent representations. Experimental results show that Joycent improves accentedness while preserving speaker identity compared with baseline systems. We release our code and demos at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16417
PDF
-
An Asymmetric Formula for Interval Consonance and its Relation to Harmonic Coincidence
Abstract
Euler's Gradus Suavitatis (1739) assigns a dissonance value to a musical interval p/q by the formula G(p/q) = 1 + \Omega^(p) + \Omega^(q), where \Omega^(n) = \sum_i e_i(p_i - 1) sums the weighted prime exponents of n. We propose the simpler asymmetric formula f(p/q) = p + \Omega^(q), which treats numerator and denominator differently and performs comparably on standard consonance data. We also show that, under a model in which harmonics are integer-indexed and counted uniformly up to a fixed truncation level, Gradus is equivalent to a weighted harmonic coincidence count with weights w(n) = \Omega^(n), connecting it to Galileo's earlier pulse-coincidence model (1638). The formula naturally generates a coprime integer triangle T(n,k) = n + \Omega^(k), whose rightmost diagonal gives the two-stage dissonance of the superparticular (consecutive-harmonic) intervals. The formula f admits a simple two-stage interpretation in terms of harmonic context and partial recognition, which we offer as a speculative perceptual hypothesis.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16412
PDF
-
Stabilizing Short Duration Speaker Verification through Neural Re-scoring with Hybrid Enrollment
2026-06-15 02:08:09Zhiqi Ai, Han Cheng, Shiyi Mu, Zhiyong Chen, Yongjin Zhou, Shugong XuAbstract
Short-duration speaker verification (SDSV) is crucial for personalized keyword spotting, where test utterances are typically shorter than three seconds. Limited speech duration results in unstable speaker representations and increased sensitivity to noise and phoneme variations, thereby degrading performance. To investigate this issue, we construct VoxPhrase, a large-scale SDSV corpus automatically segmented from the VoxCeleb dataset. Our analysis shows that text-dependent (TD) enrollment is constrained by duration and yields unstable speaker representations. In contrast, although text-independent (TI) enrollment introduces content mismatch, its representations become more stable as the enrollment duration increases. Accordingly, we propose a hybrid-enrollment neural re-scoring framework that combines TD and TI enrollment and performs frame-level comparison via parallel cross-attention. Experiments on VoxPhrase demonstrate consistent improvements across multiple speaker models.
Abstract (translated)
URL
https://arxiv.org/abs/2606.16115
PDF
-
Bridging the SEA Gap: An Initial Benchmark for Neural Audio Codec-Synthesized Speech Deepfakes in South-East Asian Languages
2026-06-14 18:50:14Orchid Chetia Phukan, Girish, Mohd Mujtaba Akhtar, Arun Balaji BuduruAbstract
Codecfakes (CFs) are a type of speech deepfakes generated through Audio Language Models (ALMs), with Neural Audio Codecs (NACs) forming the core mechanism for speech encoding and generation. CFs exhibit distributional characteristics that differ from vocoder-based deepfakes, causing detectors trained on vocoder data to generalize poorly to CFs detection. Although this has led to the development of CF detection benchmarks, existing resources are largely confined to English -- and to a limited extent Chinese -- leaving South-East Asian (SEA) languages unexplored. To bridge this gap, we introduce SEA-CF, the first large-scale benchmark for CF detection spanning multiple SEA languages, diverse speaker profiles, and a wide range of NAC architectures. SEA-CF is constructed by synthesizing publicly available real speech corpora. Our experiments show that state-of-the-art (SOTA) CF detectors trained on English-centric datasets fail to generalize to SEA speech due to language-specific phonetic structures, tonal variations, and rich prosodic diversity. We further conduct a comprehensive zero-shot and fine-tuned evaluation of recent SOTA ALMs on SEA-CF. Fine-tuning the ALMs improves performance, however, these are very large being impractical for real-world application due to their scale, particularly in low-resource and latency-constrained settings. To address this limitation, we propose a novel small-ALM, GARUDA tailored for CF detection, which delivers strong performance while remaining lightweight. Extensive evaluations demonstrate that the proposed Small-ALM outperforms strong end-to-end and ALM-based baselines, establishing a new, practical direction for robust CF detection in SEA languages and beyond.
Abstract (translated)
URL
https://arxiv.org/abs/2606.15968
PDF
