arXiv:2607. 09134v1 Announce Type: cross Abstract: Representation alignment (REPA) has been investigated to accelerate diffusion training, but we observe that regularizing intermediate representations in diffusion Transformers (DiT) may implicitly entangle latents and limit generative capacity.
arXiv:2609.13909v1 Announce Type: cross Abstract: Continuous-latent Autoregressive Diffusion Transformer (AR-DiT) models have demonstrated immense potential in zero-shot speech generation. However, t...
arXiv:2606. 09048v1 Announce Type: cross Abstract: Removing intermediate representations and separately trained decoding stages has become an important direction in generative modeling.
MeanVoiceFlow2 is a new voice conversion framework that jointly optimizes a flow-based conversion module and a computationally efficient content encoder. It is trained via conversion distillation from MeanVoiceFlow and real data reconstruction, and further enhanced with diffusion-GAN training, sample mixing, and teacher-guided conditioning augmentation. Experiments on zero-shot voice conversion show that MeanVoiceFlow2 delivers higher perceptual quality and about nine times faster inference than its predecessor while preserving speaker similarity.
arXiv:2603. 01006v3 Announce Type: replace-cross Abstract: REPresentation Alignment (REPA) improves the training of generative flow models by aligning intermediate hidden states with pretrained teacher features, but its effectiveness in token-conditioned audio Flow Matching critically depends on the choice of supervised layers, which is typically made heuristically based on the depth.
arXiv:2601. 09239v5 Announce Type: replace-cross Abstract: Speech tokenizers are a key building block of fully discrete Speech LLMs.
Flow-matching approaches to voice conversion (VC) have gained attention owing to their high speech quality and strong speaker similarity. Among them, one-step models such as MeanVoiceFlow are particul...
arXiv:2606. 20457v1 Announce Type: cross Abstract: Classifier guidance is a way to control diffusion generation by using a noise-conditioned classifier to steer the sampling process toward a target class.
The paper introduces Alignment-Free Text‑Audiobox (Text‑AB), a unified diffusion‑based framework that performs high‑quality voice dubbing and full‑duplex dialogue synthesis without requiring forced alignment. Text‑AB uses a latent diffusion model with DAC‑VAE features, achieving over 10× compression compared to prior EnCodec representations, and learns text‑speech alignment via cross‑attention. The authors pretrain a 3B‑parameter model on 480k hours of monolingual speech and fine‑tune it for cross‑lingual dubbing, full‑duplex dialogue, and emotional dialogue, reporting significant improvements in prosody, voice similarity, naturalness, and emotional expressivity over existing internal systems.
Conditional Flow Matching models for text‑to‑speech often produce incoherent frequency evolution during inference. The authors propose a training‑free, frequency‑selective boosting strategy that uses the Discrete Wavelet Transform to dynamically modulate mel‑spectrogram sub‑bands during ODE integration, penalizing aggressive low‑frequency growth while boosting lagging high‑frequency details. Across multiple architectures, this method reduces the number of function evaluations from 32 to 26 and improves Frechet Audio Distance by up to 61% without harming mean opinion scores, speaker similarity, or intelligibility.
arXiv:2606. 24745v1 Announce Type: cross Abstract: Generative models, particularly diffusion and score-based approaches, have recently achieved strong performance in speech enhancement, but their iterative sampling process limits real-time deployment.
The paper introduces Corrective Forcing (CoF), a post‑training method that aligns diffusion and flow generative models for speech enhancement by training them on self‑generated rollout states. CoF corrects predictions toward ground truth under dynamic sampling schedules and regularizes local evolution with counterfactual transitions, applying a unified objective across both model types. Experiments on SB‑VE and OT‑CFM show improved perceptual quality, reconstruction fidelity, and robustness to varying sampling steps.
arXiv:2508. 07048v2 Announce Type: replace-cross Abstract: Autoregressive (AR) encoder-decoder models dominate high-quality multilingual ASR, but their left-to-right decoders make inference latency scale with transcript length.