Pith. sign in

REVIEW 50 cited by

NaturalSpeech 3: Zero-Shot Speech Synthesis with Factorized Codec and Diffusion Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2403.03100 v3 pith:Y7EZTROB submitted 2024-03-05 eess.AS cs.AIcs.CLcs.LGcs.SD

classification eess.AScs.AIcs.CLcs.LGcs.SD
keywords speechfactorizednaturalspeechprosodyattributesdiffusiongeneratemodels
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

While recent large-scale text-to-speech (TTS) models have achieved significant progress, they still fall short in speech quality, similarity, and prosody. Considering speech intricately encompasses various attributes (e.g., content, prosody, timbre, and acoustic details) that pose significant challenges for generation, a natural idea is to factorize speech into individual subspaces representing different attributes and generate them individually. Motivated by it, we propose NaturalSpeech 3, a TTS system with novel factorized diffusion models to generate natural speech in a zero-shot way. Specifically, 1) we design a neural codec with factorized vector quantization (FVQ) to disentangle speech waveform into subspaces of content, prosody, timbre, and acoustic details; 2) we propose a factorized diffusion model to generate attributes in each subspace following its corresponding prompt. With this factorization design, NaturalSpeech 3 can effectively and efficiently model intricate speech with disentangled subspaces in a divide-and-conquer way. Experiments show that NaturalSpeech 3 outperforms the state-of-the-art TTS systems on quality, similarity, prosody, and intelligibility, and achieves on-par quality with human recordings. Furthermore, we achieve better performance by scaling to 1B parameters and 200K hours of training data.

Discussion (0). Sign in to comment.

Forward citations

Cited by 50 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. WavTTS: Towards High-Quality Zero-Shot TTS via Direct Raw Waveform Modeling

    eess.AS 2026-06 unverdicted novelty 8.0 of 10

    WavTTS is the first raw-waveform diffusion TTS model using DiT flow matching and multi-scale mel supervision that approaches SOTA latent zero-shot performance while beating prior end-to-end models.

  2. FlexiSLM: A Dynamic and Controllable Frame Rate Spoken Language Model

    cs.SD 2026-06 unverdicted novelty 7.0 of 10

    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 l...

  3. MAVIN: Multi-Shot Audio-Visual Generation with Customized Narrative Control

    cs.CV 2026-06 conditional novelty 7.0 of 10

    MAVIN is a dual-tower diffusion framework that produces temporally aligned multi-shot audio-visual content from hierarchical captions with optional multi-identity image and audio references.

  4. AudioCALM: Continuous Autoregressive Language Modeling for Universal Audio Generation

    eess.AS 2026-06 unverdicted novelty 7.0 of 10

    AudioCALM presents a continuous autoregressive framework with flow-matching prediction and A-MoME architecture that unifies speech, sound, and music generation while matching modality-specific state-of-the-art performance.

  5. Bagpiper-TTS: Natural Language Guided Universal Speech Synthesis

    cs.CL 2026-06 unverdicted novelty 7.0 of 10

    Bagpiper-TTS uses natural language prompts and intent reasoning to derive rich captions that guide a single model for universal speech synthesis across classical TTS, multi-talker, singing, and role-play tasks.

  6. AffectCodec: Emotion-Preserving Neural Speech Codec for Expressive Speech Modeling

    cs.SD 2026-05 unverdicted novelty 7.0 of 10

    AffectCodec is an emotion-guided neural speech codec that preserves emotional cues during quantization while maintaining semantic fidelity and prosodic naturalness.

  7. PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    PairAlign learns compact variable-length token sequences for audio via self-alignment on paired content-preserving views, achieving 55% fewer archive tokens than VQ while preserving edit-distance retrieval at 12.71 tokens/s.

  8. PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    PairAlign learns compact audio token sequences via self-alignment of paired content views using an autoregressive decoder, achieving strong cross-view consistency and edit-distance preservation while reducing token co...

  9. Minimizing Modality Gap from the Input Side: Your Speech LLM Can Be a Prosody-Aware Text LLM

    cs.CL 2026-05 unverdicted novelty 7.0 of 10

    TextPro-SLM minimizes the speech-text modality gap from the input side via a prosody-aware unified encoder, delivering the lowest gap and strong performance at 3B/7B scales with only ~1000 hours of audio.

  10. JUST-DUB-IT: Video Dubbing via Joint Audio-Visual Diffusion

    cs.GR 2026-01 unverdicted novelty 7.0 of 10

    JUST-DUB-IT adapts a joint audio-visual diffusion model via LoRA to generate high-quality dubbed videos with translated audio and lip-synced facial motion.

  11. Moshi: a speech-text foundation model for real-time dialogue

    eess.AS 2024-09 accept novelty 7.0 of 10

    Moshi is the first real-time full-duplex spoken large language model that casts dialogue as speech-to-speech generation using parallel audio streams and an inner monologue of time-aligned text tokens.

  12. DELTA-TTS: Adapting Autoregressive Model into Diffusion Language Model for Text-to-Speech

    eess.AS 2026-07 conditional novelty 6.5 of 10

    A LoRA-plus-convolution adaptation converts an AR TTS backbone into a confidence-ordered discrete diffusion model that improves WER and speed on limited data.

  13. Stable Autoregressive Speech Generation with Low-Frame-Rate High-Dimensional Continuous Tokens

    eess.AS 2026-07 conditional novelty 6.0 of 10

    Autoregressive TTS from 8-Hz, 768-dimensional continuous tokens works when the tokenizer shapes its latent space with a low-dimensional core and an energy hierarchy, and the generator separates guidance into local, se...

  14. StellarTTS: Sparse Temporal Embedding for Low-Latency and Robust Speech Synthesis

    cs.SD 2026-07 conditional novelty 6.0 of 10

    A mobile-oriented 83M-parameter masked transformer with sparse phone-anchored temporal embeddings achieves RTF 0.08 and lower WER than MaskGCT/F5-TTS on Seed-TTS test sets.

  15. AuEmoChat: Authentic Emotion Understanding and Rendering for Conversational Speech Synthesis

    cs.SD 2026-07 conditional novelty 6.0 of 10

    AuEmoChat's learned 1,000-code emotion token space, combined with emotion-guided token merging and classifier-guided flow matching, yields higher naturalness and emotion scores than four CSS baselines on NCSSD-EmCap.

  16. AuEmoChat: Authentic Emotion Understanding and Rendering for Conversational Speech Synthesis

    cs.SD 2026-07 conditional novelty 6.0 of 10

    AuEmoChat improves conversational speech synthesis by learning a discrete 1000-code authentic emotion token space, merging redundant dialogue context, and using emotion-guided flow matching to render speech.

  17. TokAN: Accent Normalization Using Self-Supervised Speech Tokens

    cs.SD 2026-07 accept novelty 6.0 of 10

    TokAN maps L2 speech tokens to L1-like tokens via an autoregressive converter plus GRPO rewards, cutting WER to 9.23% on seven English accents without natural parallel L1-L2 recordings.

  18. FineCombo-TTS: Collaborative and Precise Controllable Speech Synthesis Using Text Descriptions and Reference Speech

    cs.SD 2026-06 unverdicted novelty 6.0 of 10

    FineCombo-TTS learns a unified acoustic representation with a CFM-based Speech Variance Predictor for flexible precise TTS control from reference audio and text descriptions, supported by the new FineEdit paired dataset.

  19. DDPO-VC: Speaker De-Identification via Diffusion Denoising Policy Optimization

    eess.AS 2026-06 unverdicted novelty 6.0 of 10

    DDPO-VC applies diffusion denoising policy optimization with dual-teacher rewards to improve speaker de-identification while preserving cognitive utility on dementia speech benchmarks.

  20. Self-Guidance: Enhancing Neural Codecs via Decoder Manifold Alignment

    cs.SD 2026-06 unverdicted novelty 6.0 of 10

    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.

  21. EmoInstruct-TTS: Dual-Path Instruction-Guided Emotional Speech Synthesis

    cs.CL 2026-06 unverdicted novelty 6.0 of 10

    EmoInstruct-TTS uses Emotion2embed and an Instruction-Conditioned Emotion Flow Model (ICE-Flow) to generate acoustically grounded emotion representations from free-form instructions and integrate them into an LLM-base...

  22. Can We Hear from Events? Generating Speech from Event Camera

    cs.MM 2026-05 unverdicted novelty 6.0 of 10

    EventSpeech is a text-conditioned neural framework that uses neuromorphic event cameras and a new EVT-SPK benchmark to generate expressive speech, claiming to outperform RGB baselines by preserving fine-grained emotio...

  23. SemaVoice: Semantic-Aware Continuous Autoregressive Speech Synthesis

    eess.AS 2026-05 unverdicted novelty 6.0 of 10

    SemaVoice adds SFM-guided alignment to refine continuous speech representations in autoregressive TTS, reporting 1.71% English WER on Seed-TTS and competitiveness with open-source SOTA.

  24. Text-To-Speech with Chain-of-Details: modeling temporal dynamics in speech generation

    eess.AS 2026-04 unverdicted novelty 6.0 of 10

    Chain-of-Details (CoD) is a cascaded TTS method that explicitly models temporal coarse-to-fine dynamics with a shared decoder, achieving competitive performance using significantly fewer parameters.

  25. MimicLM: Zero-Shot Voice Imitation through Autoregressive Modeling of Pseudo-Parallel Speech Corpora

    cs.SD 2026-04 unverdicted novelty 6.0 of 10

    MimicLM achieves better naturalness in zero-shot voice imitation by autoregressively modeling pseudo-parallel data with synthetic sources and real targets, plus interleaved text-audio guidance and preference alignment.

  26. Bridging the Stability-Expressivity Gap: Synthetic Data Scaling and Preference Alignment for Low-Resource Spoken Language Models

    cs.CL 2026-04 conditional novelty 6.0 of 10

    Synthetic data for low-resource spoken language models creates a Stability-Expressivity Gap that DGSA and TDSC self-alignment close, enabling SOTA Thai TTS and first Lao zero-shot voice cloning.

  27. OmniCustom: Sync Audio-Video Customization Via Joint Audio-Video Generation Model

    cs.SD 2026-02 conditional novelty 6.0 of 10

    A zero-shot model that generates a video of a reference face speaking user-chosen text with a reference voice timbre.

  28. Qwen3-TTS Technical Report

    cs.SD 2026-01 unverdicted novelty 6.0 of 10

    Qwen3-TTS delivers state-of-the-art multilingual TTS performance with 3-second voice cloning, description control, and ultra-low-latency streaming via dual tokenizers and a dual-track LM architecture trained on over 5...

  29. UniVoice: Unifying Autoregressive ASR and Flow-Matching based TTS with Large Language Models

    eess.AS 2025-10 conditional novelty 6.0 of 10

    A single LLM can do ASR and zero-shot TTS on continuous speech features by switching between causal and bidirectional attention, reaching competitive but not state-of-the-art results.

  30. DiTReducio: A Training-Free Acceleration for DiT-Based TTS via Progressive Calibration

    cs.SD 2025-09 conditional novelty 6.0 of 10

    DiTReducio is a training-free, pattern-guided layer and branch skipping method that accelerates DiT-based TTS, reporting significant FLOP and RTF reductions with modest quality loss at tuned thresholds.

  31. DeCodec: Rethinking Audio Codecs as Universal Disentangled Representation Learners

    cs.SD 2025-09 conditional novelty 6.0 of 10

    DeCodec learns a single neural codec that disentangles speech, background sound, semantic content, and paralinguistic style into orthogonal quantized streams, enabling reconstruction, enhancement, voice conversion, AS...

  32. Entropy-based Coarse and Compressed Semantic Speech Representation Learning

    cs.CL 2025-08 conditional novelty 6.0 of 10

    Predictive entropy from a token-level speech language model finds merge boundaries, producing compressed semantic tokens that keep ASR and translation accuracy at 15 Hz while lowering latency.

  33. Next Tokens Denoising for Speech Synthesis

    cs.SD 2025-07 conditional novelty 6.0 of 10

    Dragon-FM generates speech autoregressively over two-second chunks while using flow matching inside each chunk, achieving fast synthesis at 12.5 discrete audio tokens per second.

  34. JAM-Flow: Joint Audio-Motion Synthesis with Flow Matching

    cs.CV 2025-06 unverdicted novelty 6.0 of 10

    JAM-Flow introduces a unified flow-matching model with a Multi-Modal Diffusion Transformer that jointly synthesizes facial motion and speech from text, audio, or motion inputs.

  35. NouveauVoice: Generating Novel Pseudo Speakers for Voice Anonymization

    eess.AS 2026-07 conditional novelty 5.5 of 10

    A hierarchical NVAE plug-in generates diverse pseudo-speaker embeddings that raise ASV EER above 38% on FACodec/CosyVoice2 with a controllable privacy-utility trade-off.

  36. AutoSIFT: Automatic Style Sifting for Controllable Speech Generation with Arbitrary Style Infilling

    cs.SD 2026-07 conditional novelty 5.0 of 10

    A TTS framework that replaces only text-specified style categories and infills all unspecified and residual style from a reference speech recording.

  37. AutoSIFT: Automatic Style Sifting for Controllable Speech Generation with Arbitrary Style Infilling

    cs.SD 2026-07 unverdicted novelty 5.0 of 10

    AutoSIFT disentangles text-describable style categories from residual speech styles and selectively infills only the categories the user specifies.

  38. AugCodec: A Low-Bitrate Disentangled Neural Speech Codec via Data Augmentation

    cs.SD 2026-06 unverdicted novelty 5.0 of 10

    AugCodec disentangles speech into semantic, speaker, and prosody tokens via tailored data augmentations, achieving 12.5 Hz operation with three streams and outperforming prior codecs on LibriSpeech reconstruction and ...

  39. Which Speech Representation Better Matches Text-Native Reasoning? A Study of Speech-Text Alignment on Frame Rate and Representation

    eess.AS 2026-06 unverdicted novelty 5.0 of 10

    Empirical sweep finds 4.17 Hz frame rate plus intermediate-layer alignment optimal for speech QA under frozen text LLM backbone.

  40. UniVoice: A Unified Model for Speech and Singing Voice Generation

    cs.SD 2026-06 unverdicted novelty 5.0 of 10

    UniVoice is a conditional flow matching model with a Diffusion Transformer backbone that unifies TTS and SVS via modality-specific encoders and a null melody token for speech, achieving 5.26% speech PER and 16.22% sin...

  41. Minimizing Modality Gap from the Input Side: Your Speech LLM Can Be a Prosody-Aware Text LLM

    cs.CL 2026-05 unverdicted novelty 5.0 of 10

    TextPro-SLM reduces the speech-text modality gap by feeding an LLM backbone with synchronized text tokens and prosody embeddings from WhisperPro, achieving lowest gap scores at 3B/7B scales with roughly 1,000 hours of audio.

  42. Multimodal Large Language Model-Enabled Video Translation: A Role-Oriented Survey

    cs.CV 2026-04 accept novelty 5.0 of 10

    MLLM-enabled video translation is usefully framed as three roles—Semantic Reasoner, Expressive Performer, and Visual Synthesizer—rather than a cascade of ASR, MT, TTS, and lip-sync.

  43. Controllable Singing Style Conversion with Boundary-Aware Information Bottleneck

    cs.SD 2026-04 unverdicted novelty 5.0 of 10

    A singing voice conversion system with boundary-aware information bottleneck and high-frequency augmentation achieves the best naturalness in SVCC2025 subjective tests while using less extra data than competitors.

  44. F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching

    eess.AS 2024-10 unverdicted novelty 5.0 of 10

    F5-TTS generates natural speech from text via flow matching on DiT with simple text padding, ConvNeXt refinement, and sway sampling, trained on 100K hours multilingual data.

  45. MAVIN: Multi-Shot Audio-Visual Generation with Customized Narrative Control

    cs.CV 2026-06 unverdicted novelty 4.0 of 10

    MAVIN proposes boundary-aware attention, ID-aware propagation, a multi-agent scripting pipeline, and the MAVINSet dataset as the first framework for multi-shot audio-visual generation with narrative control, claiming ...

  46. Multimodal Group Emotion Recognition In-the-Wild Towards a Privacy-Safe Non-Individual Approach

    cs.CV 2026-05 unverdicted novelty 4.0 of 10

    Proposes cross-attention audio-video fusion and VE-MD latent-space models for group emotion recognition that avoid individual cues and report competitive performance via ablation studies on synthetic and real data.

  47. SPARCLE: SPeaker-aware Aligned Representations via Contrastive Language Embeddings

    cs.CL 2026-05 unverdicted novelty 4.0 of 10

    SPARCLE builds speaker-aware grapheme representations by contrastively aligning characters with Wav2Vec2 acoustic embeddings conditioned on speaker identity, replacing G2P for TTS and halving WER in low-resource cases.

  48. ATRIE: Adaptive Tuning for Robust Inference and Emotion in Persona-Driven Speech Synthesis

    cs.SD 2026-04 unverdicted novelty 4.0 of 10

    ATRIE disentangles timbre and prosody in a Persona-Prosody Dual-Track model distilled from a large LLM to achieve strong identity preservation (EER 0.04) and emotional speech synthesis with SOTA results on an extended...

  49. Multimodal Large Language Model-Enabled Video Translation: A Role-Oriented Survey

    cs.CV 2026-04 unverdicted novelty 4.0 of 10

    The paper offers the first focused review of MLLM-based video translation organized by a three-role taxonomy of Semantic Reasoner, Expressive Performer, and Visual Synthesizer, plus open challenges.

  50. Voice Mapping of Text-to-Speech Systems: A Metric-Based Approach for Voice Quality Assessment

    eess.AS 2026-04 unverdicted novelty 3.0 of 10

    Voice range indicates TTS model capability with VITS highest, Glow-TTS best at soft phonation, and CPPs of 7-8 dB marking natural quality while values over 10 dB sound robotic.

Pith tools