{"total":27,"items":[{"citing_arxiv_id":"2606.31729","ref_index":41,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Is Natural Always Appropriate? Investigating Naturalness and Appropriateness Across Different Domains for TTS Evaluation","primary_cat":"eess.AS","submitted_at":"2026-06-30T14:28:41+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Appropriateness of TTS varies independently across domains while naturalness scores penalize stylized speech and reward spontaneity.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.31247","ref_index":240,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"FlexiSLM: A Dynamic and Controllable Frame Rate Spoken Language Model","primary_cat":"cs.SD","submitted_at":"2026-06-30T07:24:10+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"FlexiSLM is the first spoken language model supporting dynamic and controllable frame rates on speech input and output, outperforming fixed-rate 7B models at high quality and enabling faster inference at lower rates like 6.25 Hz.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.28568","ref_index":2,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"KM-Speaker: Keypoint-Based Style Control for High-Quality Speech-Driven 3D Facial Animation and Dialogue Localization","primary_cat":"cs.CV","submitted_at":"2026-06-26T19:48:44+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"KM-Speaker introduces a keypoint-conditioned generative framework for speech-driven 3D facial animation offering global style guidance and frame-level temporal control via disentangled lip and upper-face dynamics.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.25424","ref_index":31,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Adaptive Oscillatory Inductive Bias for Modeling Sharp Prosodic Dynamics in Diffusion-Based TTS","primary_cat":"eess.AS","submitted_at":"2026-06-24T05:32:20+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"OscillaTTS adds an adaptive oscillatory nonlinearity with linear bypass to diffusion TTS decoders for improved modeling of abrupt prosodic dynamics, reporting gains on LJSpeech and Emotional Speech Dataset.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.21237","ref_index":14,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"OpenWER: Improving Cross-Lingual ASR Evaluation and Enabling Token-Based Accuracy Metrics","primary_cat":"cs.CL","submitted_at":"2026-06-19T09:00:48+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"OpenWER improves cross-lingual WER robustness via language-specific normalization, compound word detection, and token-based Levenshtein alignment, reporting up to 25% absolute reductions across 52 languages.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.20218","ref_index":21,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Zero-VC: Zero-Lookahead Streaming Voice Conversion via Speaker Anonymization","primary_cat":"cs.SD","submitted_at":"2026-06-18T13:32:57+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Zero-VC applies speaker anonymization as a perturbation to achieve strictly causal zero-lookahead streaming voice conversion by balancing timbre leakage against prosodic utility.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.18094","ref_index":11,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Next-Turn: Duration-Aware Streaming Endpoint Detection via Time-to-Next-Speech-Onset Prediction","primary_cat":"cs.SD","submitted_at":"2026-06-16T15:58:36+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Next-Turn introduces time-to-next-speech-onset prediction for duration-aware streaming endpoint detection, reporting a 25.9% improvement in accuracy within 320 ms.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.17742","ref_index":5,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"BrainWorld: A Structural-Prior-Conditioned Generative Model for Whole-Brain 4D fMRI Dynamics","primary_cat":"cs.CV","submitted_at":"2026-06-16T10:03:47+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"BrainWorld is a structural-prior-conditioned generative model that produces stable whole-brain 4D fMRI trajectories up to 400 frames, augments downstream tasks, and learns transferable multimodal representations across 22 datasets.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.12940","ref_index":93,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Self-Guidance: Enhancing Neural Codecs via Decoder Manifold Alignment","primary_cat":"cs.SD","submitted_at":"2026-06-11T06:06:02+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Self-guidance adds a lightweight feature-mapping loss to align decoder manifolds in VQ-VAE speech codecs, raising reconstruction metrics and allowing 4x codebook reduction with no fidelity loss.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11542","ref_index":30,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Pretrained self-supervised speech models can recognize unseen consonants","primary_cat":"cs.CL","submitted_at":"2026-06-10T01:07:32+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Fine-tuned Wav2Vec2 and HuBERT models recognize click consonants more accurately than non-clicks in G|ui and West !Xoon data.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11429","ref_index":9,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Gumbel-BEARD: Automatic Layer Selection for Self-Supervised Adaptation of Whisper in Low-Resource Domains","primary_cat":"eess.AS","submitted_at":"2026-06-09T20:27:59+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Gumbel-BEARD automates Whisper layer selection with Gumbel-Softmax and BEST-RQ for self-supervised domain adaptation, matching fully supervised performance on 10h vs 133h data and setting new SOTA WERs on MyST and OGI datasets.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.10758","ref_index":18,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Anchoring the Unknown: Open-Set Model Attribution via Proxy-Anchor Learning","primary_cat":"eess.AS","submitted_at":"2026-06-09T12:10:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Proxy-Anchor metric learning on Wav2Vec2-BERT embeddings with architecture merging achieves 99.76% closed-set accuracy and 2.04% FPR@95 OOD detection on MLAAD v9, doubling prior OOD accuracy on v5 splits.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.06743","ref_index":13,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"HybridCodec: Fast Dual-Stream, Semantically Enhanced Neural Audio Codec","primary_cat":"cs.SD","submitted_at":"2026-06-04T21:57:18+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"HybridCodec unifies SSL distillation and dual-stream design in a neural audio codec for improved semantic specialization, competitive reconstruction, and faster inference.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.06740","ref_index":10,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Multilingual Multi-Speaker Unit Vocoders: A Systematic Analysis of Discrete Speech Representations","primary_cat":"cs.SD","submitted_at":"2026-06-04T21:54:56+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Empirical study of unit vocoders finds cluster size controls intelligibility via phonetic discriminability and speaker conditioning is required to avoid collapse in multilingual multi-speaker settings.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.05121","ref_index":4,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Audio Interaction Model","primary_cat":"cs.SD","submitted_at":"2026-06-03T17:26:11+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Audio-Interaction unifies offline and online audio tasks into one streaming model via the SoundFlow framework and a new 2.6M-item streaming corpus, enabling real-time instruction following and proactive responses.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.03241","ref_index":14,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Benchmarking Speech-to-Speech Translation Models","primary_cat":"cs.CL","submitted_at":"2026-06-02T07:01:33+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"COMPASS is a new reproducible benchmarking framework for S2ST that deploys 46 metrics on 1248 configurations, shows single-metric rankings mislead, reduces to 10 metrics per direction, and finds domain-specific metrics better match human judgments than standalone MOS predictors.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.31173","ref_index":28,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"MindVoice: Reconstructing Intelligible Speech from Non-invasive Neural Signals with Pretrained Priors","primary_cat":"cs.SD","submitted_at":"2026-05-29T11:38:38+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"MindVoice disentangles neural-to-speech reconstruction into semantic and acoustic pathways using pretrained priors, then fuses them with speech generation models to produce intelligible output from non-invasive recordings.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.11098","ref_index":22,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"AffectCodec: Emotion-Preserving Neural Speech Codec for Expressive Speech Modeling","primary_cat":"cs.SD","submitted_at":"2026-05-11T18:04:33+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"AffectCodec is an emotion-guided neural speech codec that preserves emotional cues during quantization while maintaining semantic fidelity and prosodic naturalness.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.10084","ref_index":25,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"PoDAR: Power-Disentangled Audio Representation for Generative Modeling","primary_cat":"eess.AS","submitted_at":"2026-05-11T07:05:11+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"PoDAR disentangles audio signal power from semantic content in latents using power augmentation and consistency objectives, yielding 2x faster convergence and gains of 0.055 speaker similarity and 0.22 UTMOS when applied to Stable Audio VAE with F5-TTS.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Although these advancements were primarily developed for the image domain, there has been parallel progress on the audio front. In the development of the Moshi codec [10], the authors align the primary codebook latents with a pre-trained WavLM encoder [24] and similarly, the DualCodec framework [11] aligns its representations with a pre-trained w2v-BERT-2.0 model [25]. Classifier-free guidanceClassifier-Free Guidance (CFG) [ 15] is a fundamental component of modern generative models, since it not only improves conditional adherence, but also generation 3 quality [26]. It achieves this by extrapolating from an unconditional baseline toward a conditional prediction: vcfg =v 0 +w(v cond −v 0).(2) where v0 is the unconditional prediction, vcond the conditional prediction, and the scalar w controls"},{"citing_arxiv_id":"2604.26347","ref_index":14,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"The False Resonance: A Critical Examination of Emotion Embedding Similarity for Speech Generation Evaluation","primary_cat":"eess.AS","submitted_at":"2026-04-29T06:59:48+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Emotion embedding similarities are unsuitable for zero-shot evaluation of emotional expressiveness in speech generation due to confounding by non-emotional acoustic features.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.22467","ref_index":10,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"DM-ASR: Diarization-aware Multi-speaker ASR with Large Language Models","primary_cat":"eess.AS","submitted_at":"2026-04-24T11:43:25+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"DM-ASR reformulates multi-speaker ASR as multi-turn dialogue generation conditioned on diarization results, achieving competitive benchmark performance with relatively small models and limited data.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.17435","ref_index":18,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"MoVE: Translating Laughter and Tears via Mixture of Vocalization Experts in Speech-to-Speech Translation","primary_cat":"cs.CL","submitted_at":"2026-04-19T13:34:52+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"MoVE uses specialized LoRA expert adapters and a soft router to translate non-verbal vocalizations in S2ST, reproducing them in 76% of cases versus at most 14% for baselines while scoring highest on naturalness and emotional fidelity.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.16287","ref_index":6,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"NaijaS2ST: A Multi-Accent Benchmark for Speech-to-Speech Translation in Low-Resource Nigerian Languages","primary_cat":"cs.SD","submitted_at":"2026-04-17T17:49:46+00:00","verdict":null,"verdict_confidence":null,"novelty_score":null,"formal_verification":null,"one_line_summary":null,"context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.09344","ref_index":1,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"DialogueSidon: Recovering Full-Duplex Dialogue Tracks from In-the-Wild Dialogue Audio","primary_cat":"cs.SD","submitted_at":"2026-04-10T14:16:34+00:00","verdict":null,"verdict_confidence":null,"novelty_score":null,"formal_verification":null,"one_line_summary":null,"context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.05830","ref_index":9,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"\"OK Aura, Be Fair With Me\": Demographics-Agnostic Training for Bias Mitigation in Wake-up Word Detection","primary_cat":"cs.CL","submitted_at":"2026-04-07T13:02:04+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Demographics-agnostic training with augmentation and distillation reduces predictive disparity in wake-up word detection by 40-84% across demographic groups.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.04598","ref_index":5,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Benchmarking Multilingual Speech Models on Pashto: Zero-Shot ASR, Script Failure, and Cross-Domain Evaluation","primary_cat":"cs.CL","submitted_at":"2026-04-06T11:23:42+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Multilingual ASR models show 39.7-297% zero-shot WER on Pashto public data, Whisper models output correct script in under 0.8% of cases, and fine-tuned models degrade to 32.5-59% WER on out-of-domain sets.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2504.01919","ref_index":105,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Bridging the Linguistic Divide: A Survey on Leveraging Large Language Models for Machine Translation","primary_cat":"cs.CL","submitted_at":"2025-04-02T17:26:40+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":3.0,"formal_verification":"none","one_line_summary":"A literature survey that organizes prompting, fine-tuning, preference optimization, and context-aware techniques for LLM-based machine translation with emphasis on low-resource languages.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"given sufficient high-quality parallel data and careful multilingual balancing, conven- tional MLE-based fine-tuning can deliver strong and stable translation performance across dozens of language pairs without reliance on preference modeling or reinforce- ment learning, while also serving as a strong teacher signal for cross-modal transfer. Hunyuan-MT [105] treats SFT as the primary driver of translation quality, applying large-scale, multi-stage parallel-data SFT (millions of pairs, heavily filtered) before RL refinement. ALMA [106] relies entirely on SFT for adaptation, using a minimal- ist two-stage design: monolingual SFT for multilingual competence followed by small, 20 high-quality parallel SFT for translation induction."}],"limit":50,"offset":0}