Research map / Probabilistic and scientific modeling / Variational and generative sampling
Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use end-to-end VITS instead of the strongest two-stage baseline | LJ Speech TTS · LJ Speech test | MOS: 4.32 ($\pm$0.07) -> 4.43 ($\pm$0.06) | Helped |
| use end-to-end VITS instead of the strongest two-stage baseline | VCTK multi-speaker TTS · VCTK test | MOS: 3.82 ($\pm$0.07) -> 4.38 ($\pm$0.06) | Helped |
| add a stochastic duration predictor rather than a deterministic duration predictor (ablation) | VITS · LJ Speech test | MOS: 4.39 ($\pm$0.06) -> 4.43 ($\pm$0.06) | Helped |
| remove the normalizing flow from the prior encoder (ablation) | VITS · LJ Speech ablation | MOS: 4.50 ($\pm$0.06) -> 2.98 ($\pm$0.08) | Hurt |
| replace linear-scale spectrogram posterior input with mel-spectrogram input (ablation) | VITS · LJ Speech ablation | MOS: 4.50 ($\pm$0.06) -> 4.31 ($\pm$0.08) | Hurt |
| use a deterministic duration predictor rather than the stochastic duration predictor (ablation) | VITS · LJ Speech | Speed (kHz): 1480.15 -> 2005.03 | Helped |
Closest papers by meaning
- End-to-end Adversarial Text-to-Speech (ICLR 2021 · similarity 0.90)
- Bidirectional Variational Inference for Non-Autoregressive Text-to-Speech (ICLR 2021 · similarity 0.89)
- High Fidelity Speech Synthesis with Adversarial Networks (ICLR 2020 · similarity 0.88)
- NaturalSpeech 2: Latent Diffusion Models are Natural and Zero-Shot Speech and Singing Synthesizers (ICLR 2024 · similarity 0.87)
- A Variational Framework for Improving Naturalness in Generative Spoken Language Models (ICML 2025 · similarity 0.87)
- Continuous Autoregressive Modeling with Stochastic Monotonic Alignment for Speech Synthesis (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.