AI Voice Conversion
AI Voice Conversion 是 osFoundry 社区目录中的应用。从某说话者的样本训练一个声音模型,然后将任意音频转换为该说话者的声音。对歌唱友好(保留音高 + 气息)。基于 so-vits-svc 构建。**⚠️ 训练强烈建议使用 GPU(12+ GB VRAM);推理 CPU 可用**。
详情
- 工作区: osfoundry
- 分类: COMMUNICATION
- 价格: Free
- 访问权限: Community
功能
- Train a voice model from samples of a speaker
- Then convert any audio to that speaker's voice
文档
文档由上游项目以英文维护。
# AI Voice Conversion
Train a voice model from samples + convert audio to that voice, powered by so-vits-svc.
## ⚠️ Heavy + GPU-bound for training
- **Training a new voice model**: 12+ GB VRAM, ~hours per model. **CPU is impractical**.
- **Inference (using an existing model)**: CPU-workable at slower-than-real-time speeds.
## Features
- Singing voice conversion (preserves pitch + breath dynamics)
- Train from 30 min - 2 hours of clean speech / vocal samples
- Hubert + ContentVec acoustic feature encoders
- f0 modeling (pitch extraction + reconstruction)
- Gradio UI for inference
## Ethical notes
Voice cloning has obvious abuse potential. Don't train models of voices you don't have explicit permission to clone. Don't release synthesised audio of real people without consent.
## Packaging
Python + Gradio + so-vits-svc from upstream. Models + training data at `/data`.
## CONFIRM-AT-BUILD
so-vits-svc has many forks. The base repo (svc-develop-team) is the canonical one but its CLI evolves quickly — pin a specific commit.
如何在 osFoundry 中使用 AI Voice Conversion
一键将 AI Voice Conversion 安装到您的工作区,然后在 osStudio 中将其分支,针对您的技术栈自定义提示词、工具或配置。工作区中的任何人都可以接续您的工作继续推进。
社区中的其他应用
- 客户关系管理 — 客户关系管理工具,支持联系人、交易和销售管道跟踪。
- Kanban Board — Trello 风格的看板与项目板,含卡片、面板、日历与表格视图以及每面板属性。基于 Focalboard(独立个人服务器)构建。在持久卷上内嵌 SQLite。
- 服务台 — 工单分流与客户支持收件箱,附带 SLA 跟踪。
- Page Builder — 可视化拖放页面生成器,支持区块、主题、SEO 与发布
- Website Builder — 带 CMS 合集、全局导航、页脚、主题与发布的多页面网站生成器
- 店面 — 电商店面,包含商品目录、购物车和结账流程。