Drama 3 Local Voice Studio

An independent local text-to-speech workspace from Plops Forge. Keep scripts, reusable voice references and completed takes together on your own computer.

Features

• Persistent drafts and generation settings.
• Reusable voice references and searchable take history.
• Offline English spelling and grammar checks, formatting with Undo, and optional local Whisper dictation.
• Choose a Windows output folder once for automatic saving of new completed takes.
• Per-take Save as with safe filenames and collision handling. Internal history keeps its copy if external saving fails.

Download and install

Download local-voice-studio-0.1.1-source.zip from this page and extract it into a writable folder. Read the included RELEASE.md for Windows prerequisites, the setup command and startup instructions. This is source requiring setup, not a bundled executable.

The documented configuration requires Python 3.10+, Node.js 22.12+ and npm during setup, compatible NVIDIA/CUDA/Visual C++ prerequisites, a separately obtained and appropriately licensed Drama model, and a trusted audio.cpp engine. Model weights, recorded voices, private data and engine binaries are not included.

Follow RELEASE.md to configure the model and engine. The standalone launcher does not require Pinokio. Start with a short test take and verify your save destination before a longer session. Preserve your existing drafts, history and configuration when upgrading.

Validation and limits

The public release records 54 passing tests. Existing-engine setup and synthetic generation were tested. Actual native picker selection remains user-operated and pending confirmation; a complete clean CUDA/toolchain installation has not been validated. Read VALIDATION.md for the full scope. Screenshots show the studio and voice library from an isolated demonstration with synthetic text.

Licensing

The UI source is Apache-2.0. Model weights and the engine have separate licenses. The UI license does not grant model rights or unrestricted commercial text-to-speech rights. This is an independent interface, not an official Fish Audio product.

Public v0.1.1 release · Source repository · Written setup/workflow guide

The attached ZIP is byte-identical to the public GitHub v0.1.1 asset. Its SHA-256 is d6e18a6553dca068d969b7fd2724c4a0489d1b6d98537dcb65ee4b63ce8be81d. Newer unpublished packages are not included.

Tips are optional support for Plops; downloading does not require a tip.