Abstract
A living systematic review of synthetic speech, covering text-to-speech (TTS), voice conversion (VC), and spoken conversational agents. The wiki currently contains 640 paper pages and tracks 23 concepts.
The wiki turns a growing paper corpus into three complementary ways to read the research:
- Concept Overviews give a concise account of what matters now, for readers scanning the field.
- Concept In Depth pages explain what the reviewed research collectively shows, where findings disagree, and which conclusions remain uncertain.
- Structured claim YAML preserves the complete machine-readable links among claims, evidence, method families, and papers.
Concept conclusions describe the literature integrated into this corpus. They are not claims of exhaustive coverage of the field.
Warning
Work in progress. Concept integration and synthesis currently cover research through Q3 2025. The paper catalog may contain later work that has been ingested but not yet incorporated into concept conclusions. Coverage will advance chronologically until it catches up with current research, then continue on a rolling basis.
→ Choose how to explore the wiki
Read the Integrated Concepts
Five concepts have completed integration and are available at both reading depths.
| Concept | Overview | In Depth | Papers reviewed |
|---|---|---|---|
| Flow Matching | Overview | In Depth | 97 |
| Speech-to-Speech Systems | Overview | In Depth | 60 |
| Evaluation Metrics | Overview | In Depth | 285 |
| RLHF for Speech | Overview | In Depth | 29 |
| Disentanglement | Overview | In Depth | 100 |
Paper counts overlap because one paper can inform several concepts.
→ See all 23 concepts and their integration status
Cross-Concept View
The Field Overview connects findings across the five completed concept integrations. It highlights shared design pressures in generation, control, alignment, interaction, and evaluation while keeping its conclusions bounded to the reviewed corpus.
Browse the Papers
The corpus contains 640 paper pages from arXiv, Interspeech, ACL, EMNLP, NAACL, ICLR, NeurIPS, ASRU, workshops, and industry technical reports.
Project Status
The remaining 18 concepts are awaiting their Q3 2025 integration pass. Their existing pages are navigation placeholders rather than current research syntheses.
After Q3 integration is complete across the concept catalog, the project will continue through Q4 2025 and subsequent periods until the synthesis layer catches up with the live research corpus. Broader analysis is planned alongside that expansion:
- Time-based reports will compare how methods, evidence, and open questions change between periods such as Q3 and Q4 2025.
- Venue reports will synthesize the research represented at venues such as Interspeech, NeurIPS, ICLR, and ACL when coverage is sufficient for a meaningful analysis.
- Citation analysis will track the most-cited and most influential papers in the corpus, including changes in their relative position over time.