voicebox Releases

Fixed server crashes related to BBR/Reno congestion control and iptables lock contention. Added an environment variable to specify firewall backends.

Mar 31, 2026

Updated `remark-cjk-friendly` and `remark-cjk-friendly-gfm-strikethrough` dependencies to v2.x. The upgrade includes dropping Node.js 16 support, which has no impact as Streamdown already requires Node.js ≥18.

Mar 17, 2026

Added new props for code blocks (`lineNumbers` and `meta`) and improved streaming animations. Also fixed CSV downloads for Excel compatibility and unnecessary re-renders during streaming updates.

Mar 17, 2026

Added opt-in KaTeX completion for formulas and resolved false strikethrough rendering issues with single tildes.

Mar 17, 2026

A patch release for the Streamdown code plugin that improves stability by handling unknown or truncated code block language identifiers gracefully, preventing the Shiki highlighter from throwing errors.

Mar 17, 2026

Major backend refactor into a modular architecture, overhauls settings UI into routed sub-pages, and migrates documentation to Fumadocs.

Mar 17, 2026

Fixes PyInstaller bundling issues that prevented model downloads, loading, and progress tracking from working in production builds.

Mar 15, 2026

Fixes model support in bundled builds and adds CUDA binary auto-update functionality.

Mar 15, 2026

A major overhaul introducing multi-engine TTS support, post-processing audio effects, and enhanced cross-platform capabilities including Docker and Windows support.

Mar 15, 2026

This is a post-release patch for version 0.27.0. The provided text only contains a link to the full changelog and does not list specific changes or fixes.

Mar 14, 2026

Mar 8, 2026

This minor release introduces customization options for syntax highlighting by adding support for custom Shiki themes through the createCodePlugin configuration.

Mar 5, 2026

This release focuses on stability and reliability improvements across multiple areas including model selection, MLX bundling for Apple Silicon, file handling, media device security, and UX enhancements.

Feb 23, 2026

Feb 9, 2026

Major UX overhaul for model downloads with real-time progress tracking, better error handling, and improved health checks. Enhanced update notifications and GPU information reporting.

Jan 31, 2026

Quick patch release fixing transcription issues on MLX backend and resolving model download progress display problems.

Jan 30, 2026

Major performance update introducing MLX backend for Apple Silicon, achieving massive speed improvements from ~20s to 2-3s per generation. Added dynamic backend selection between MLX and PyTorch.

Jan 30, 2026

User experience improvements focusing on voice profile creation with draft saving, Stories editor enhancements, and generation box UI refinements.

Jan 30, 2026

Bug fixes addressing critical model download timeouts and cross-platform cache path issues.

Jan 29, 2026

Audio editing and playback enhancements for the Story Editor.

Jan 29, 2026