VoiceStudio

Introduces Studio's Convert method, a synchronized audiobook player, batch queue watch folders, and significant UI refinements for dubbing and engine management across Windows, macOS, and Linux.

Version 0.5.1 introduces VoiceStudio as a local speech platform with new API transports, improved dubbing tools, expanded hardware support for Apple Silicon and AMD GPUs, and a simplified one-command installation process.

OmniVoice-Studio is renamed to VoiceStudio, introducing a new Model Catalogue, remote GPU sharing via join codes, enhanced server security, and a redesigned Dub workspace.

Improved update notifications, prevented data loss during updates, fixed model download repairs, and corrected localization errors across multiple languages.

This update fixes AMD GPU acceleration, improves error reporting, adds MCP host allowlist support, and resolves various stability and compatibility issues across Windows and macOS.

Introduces comprehensive audiobook tools with multi-voice casting, a designed-voice Gallery, Paste Translation for dubbing, and significant performance and security optimizations.

This release significantly improves dubbing quality with stereo audio, better timing, and consistent voice matching, while introducing opt-in analytics and critical memory and performance fixes for 16GB machines.

This update fixes major memory leaks and OOM crashes by optimizing model unloading and introduces a comprehensive new Storage settings menu for granular data management.

Added an in-app uninstall feature, fixed a race condition causing incorrect backend connection errors, and improved cleanup of log folders on Linux and Windows.

This update improves error reporting accuracy, introduces streaming audio playback, adds dedicated uninstallers for all platforms, and optimizes the first-run wizard experience.

Introduces one-click installation for IndexTTS-2, automated Hugging Face endpoint detection for restricted networks, and improved settings concurrency and test coverage.

This release bundles FFmpeg and yt-dlp, redesigns the Engines and Models pages, fixes a critical dubbing session crash, and resolves numerous settings and scaling issues.

Introduces generation takes with restore, per-sentence caching for audiobooks, enhanced dubbing consistency and fit prediction, global text normalization, and a persistent audio mini-player.

Fixes cold-start timeouts, prevents updates from deleting manual engine installs, resolves a voice cloning performance regression, and adds new Agent Skills.

Adds visible engine pickers for TTS, ASR, and LLM in Settings, fixes Linux AppImage white-screen detection, and adds Windows installation documentation.

This release adds OpenAI-compatible transcription, fixes macOS microphone permissions, resolves Linux AppImage white-screening, and improves voice cloning and dubbing stability.

Improves engine selection consistency, fixes first-run network and SSL proxy blockers, and resolves multiple crashes and community-reported bugs across Windows, Linux, and macOS.

Improves multi-language dubbing workflows with per-language translation and caching, adds self-documenting backend crash reports, and fixes various stability issues across different platforms and network configurations.

Rebuilds dictation with live waveforms and faster commits, adds detailed LLM connection diagnostics, introduces per-feature LLM routing, and implements extensive reliability fixes for VRAM and system crashes.

This release fixes dialogue hallucinations in Cinematic/Autofit dubbing, resolves audiobook rendering crashes, ensures correct speaker counts, and prevents stale backend versions from persisting after updates.

Introduces live local dictation, a user pronunciation dictionary, a redesigned Settings hub, and critical stability fixes for backend connectivity and GPU memory crashes.

This stable release improves non-English language correctness, adds cross-platform audio and UI fixes for Linux and Android, introduces two new opt-in TTS engines, and resolves various stability and installation bugs.

Introduces a Longform suite for audiobooks and stories, engine-routing to prevent silent CPU fallback, enhanced dubbing tools, and a transition to AGPL-3.0 licensing.

Fixed a speaker diarization failure caused by PyTorch 2.6's secure unpickler by adding necessary safe-globals to the allowlist.

Fixed transcription failures on Windows with NVIDIA GPUs by introducing PyTorch Whisper as a fallback backend to eliminate cuDNN 8 dependencies.

Corrected CPU architecture reporting in Docker/web builds and improved CI checksum portability for macOS runners.

Fixed an issue where Docker UI users encountered 'Loopback origin required' errors and blank version information due to restricted admin routes.

Version 0.3.1 fixes voice-clone export crashes in Docker/browser builds, corrects version display and updater UI, and improves transcription error reporting.

Introduced a frameless OS-level dictation widget, refactored the capture component, updated Docker GPU detection, and overhauled the README.

Version 0.3.0 introduces Pro Studio for audiobook creation, extensive i18n support for 21 languages, network sharing via LAN/Tailscale, and significant stability and security improvements across Windows, macOS, and Linux.

Introduced a frameless global dictation widget, refactored the capture component into a standalone widget, and overhauled the README documentation.

Introduced a new frameless dictation widget.

This is an auto-generated release based on the commit log.

Introduces system-wide dictation with global hotkeys, an autonomous batch video dubbing pipeline, and dual-mode ASR improvements.

Introduces a WebSocket event bus for real-time UI updates, updates the default UI scale, and fixes several cross-platform stability and boot issues.

This is an auto-generated release with changes detailed in the commit log.

This is an auto-generated release with details available in the commit log.

This is an auto-generated release with changes detailed in the commit log.