Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss
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 AutoVC instead of the compared non-parallel conversion baselines | Traditional many-to-many conversion on VCTK · VCTK, seen speakers | MOS | Helped |
| Use AutoVC instead of the compared non-parallel conversion baselines | Traditional many-to-many conversion on VCTK · VCTK, seen speakers | voice similarity | Helped |
| Replace AutoVC speaker embeddings with one-hot speaker embeddings (ablation) | AutoVC · VCTK, seen speakers | MOS and voice similarity | No clear effect |
| Use AutoVC for conversion between unseen speakers | AutoVC zero-shot conversion · VCTK, unseen-to-unseen speakers | voice similarity | Mixed |
| Reduce the content bottleneck dimensions and increase temporal downsampling (ablation) | AutoVC · VCTK, 40 seen speakers | reconstruction error: 8.59 -> 34.6 | Hurt |
| Reduce the content bottleneck dimensions and increase temporal downsampling (ablation) | AutoVC · VCTK, 40 seen speakers | speaker classification accuracy: 12.0% -> 7.50% | Helped |
| Increase content dimensions, decrease temporal downsampling, and set λ to 0 (ablation) | AutoVC · VCTK, 40 seen speakers | reconstruction error: 8.59 -> 3.85 | Helped |
| Increase content dimensions, decrease temporal downsampling, and set λ to 0 (ablation) | AutoVC · VCTK, 40 seen speakers | speaker classification accuracy: 12.0% -> 70.5% | Hurt |
| Use a tuned bottleneck instead of adversarial training to disentangle speaker information | AutoVC · VCTK, 20 seen speakers | speaker classification accuracy: 45.1% -> 14.2% | Helped |
Closest papers by meaning
- StableVC: Style Controllable Zero-Shot Voice Conversion with Conditional Flow Matching (AAAI 2025 · similarity 0.88)
- Vevo: Controllable Zero-Shot Voice Imitation with Self-Supervised Disentanglement (ICLR 2025 · similarity 0.86)
- SelfVC: Voice Conversion With Iterative Refinement using Self Transformations (ICML 2024 · similarity 0.85)
- StyleSinger: Style Transfer for Out-of-Domain Singing Voice Synthesis (AAAI 2024 · similarity 0.84)
- Seeing Your Speech Style: A Novel Zero-Shot Identity-Disentanglement Face-based Voice Conversion (AAAI 2025 · similarity 0.84)
- DDDM-VC: Decoupled Denoising Diffusion Models with Disentangled Representation and Prior Mixup for Verified Robust Voice Conversion (AAAI 2024 · similarity 0.84)
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.