arXiv 2610.28940 · cs.SD · 2026-10-20
StyleStream 2.0: Fast and Controllable Streaming Voice Style Conversion
Yisi Liu, Nicholas Lee, Gopala K. Anumanchipalli
UC Berkeley
In latest · #345 Added on 2026-10-26
One-line takeaway
StyleStream 2.0 is a streaming voice style conversion system that cuts end-to-end latency on a consumer GPU to 520 ms (2x faster than 1.0) using a few-step MeanFlow model with autoregressive feedback, and adds a text interface for designing target voices.
StyleStream 2.0 是流式语音风格转换系统，用少步 MeanFlow 加自回归反馈微调，把消费级 GPU 上的端到端延迟降到 520 毫秒（比 1.0 快一倍），并加入文本设计目标音色的接口。
Read on arXiv PDF Hugging Face discussion Code Project page
Topics & disciplines
More on Efficient training & inference
- The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation ▲511
- Scaling Properties of Same-Family On-Policy Distillation ▲327
- Beyond Domain-Routed Supervision: Cross-Teacher Guarding for Multi-Teacher On-Policy Distillation ▲250
- On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics ▲205
- Beyond Teacher Assignment: Domain-Normalized Multi-Teacher On-Policy Distillation ▲181
More from UC Berkeley
About this data
Metadata from the arXiv API (verified); upvotes from Hugging Face Daily Papers; citations from Semantic Scholar. Last refreshed 2026-10-26.
The takeaway is written by AIMap from the abstract; the abstract itself is not reproduced — read the original.