LocalText2Voice v1.3.0

What's new

• Added a supported Linux-from-source workflow with a Bash launcher, Wayland detection, cross-platform Piper/FFmpeg discovery, Linux virtual environments, documentation, and tests.
• Added configurable storage for AI models, engine dependencies, voice-gallery data, and caches. Windows users can choose the location during installation and move it later from Settings.
• Added safe cross-drive migration for managed AI assets without moving projects or generated audio.
• Added detailed live package and download output for optional local-engine installation.
• Fixed audio export across different Linux filesystems (EXDEV).
• Hardened Piper status, preview, regeneration, verification, and server execution across platforms.

Windows distribution

• Clean unsigned installer for Windows 10/11 x64.
• The installer now includes a dedicated page for selecting the large AI-model storage location.
• The selected location receives a managed data folder and ownership marker; uninstall cleanup will never recursively delete an arbitrary unmarked folder.
• Silent uninstall preserves downloaded AI data.
• Includes the GUI, persistent EngineHost, MCP stdio server, FFmpeg, Piper, bundled voices/music, and a clean embedded Python runtime.
• SHA-256: a7b190e423186632a5e9b06a34e67cd3cf2b226fba81a20406071ef3bfa40d96

Linux

Linux currently runs from source. See the Linux guide or the README quick start. Debian/Ubuntu users need python3-venv and ffmpeg; Arch users need python and ffmpeg.

Validation

• 242 tests and 18 subtests passed; one known modal UI test was excluded.
• Linux runtime and executable-resolution tests passed under Ubuntu/WSL.
• A clean PyInstaller distribution contained no local config, model data, or optional engine dependencies.
• CPU/Spanish and GPU silent installations were verified, including storage.base_dir, the ownership marker, and first-run profile state.
• Installed GUI startup and EngineHost help were verified.
• Silent uninstall was verified to preserve the selected AI storage.
• Installer version, tag/tree identity, size, and SHA-256 were verified.

Thanks

Special thanks to Roman Kudryavsky (devpilgrin) for contributing and testing the Linux version. Thanks also to Evgeny Shchelakov for his detailed LocalText2Voice video.

LocalText2Voice v1.2.1

What's new

• Added a complete Russian UI, expanding LocalText2Voice to eleven interface languages. Russian is also available in the Faster Whisper selector and Windows installer.
• Added five project-source tools to the stdio MCP server, HTTP MCP transport, and local HTTP API: read_job_source, write_job_source, search_job_source, edit_job_source, and replace_job_source_text.
• Added safe agent editing for completed audiobook projects with character-based pagination, regex or literal search, Unicode offsets, and SHA-256 concurrency checks. SQLite, source.txt, and the project manifest remain synchronized.
• Added physical model-cache discovery for Chatterbox, Qwen3 TTS, OmniVoice, and Faster Whisper, plus direct asset validation for Kokoro. Missing or stale install manifests no longer hide valid downloads.
• Added Repair / Update and Reinstall / Update flows that reuse existing model files and retrieve only missing or outdated runtime/model assets.
• Improved Faster Whisper CUDA handling: if required CUDA runtime libraries cannot be loaded, review now logs the issue and transparently continues on CPU int8 instead of failing.
• Fixed automatic language handling for text normalization and restored the normalized-text editor reliably after it is re-enabled.
• Added a versioned CHANGELOG.md and updated MCP/HTTP project-editing documentation.

Windows distribution

• Clean installer for Windows 10/11 x64.
• Piper remains the lightweight default engine.
• Optional TTS engines and Faster Whisper remain on-demand downloads.
• Includes the GUI, persistent EngineHost, MCP stdio server, FFmpeg, Piper, bundled voices/music, Russian locale, and the clean embedded Python runtime.
• The installer is currently unsigned.
• SHA-256: 5be69b4b927e28ce23c87fd8ccd98ef9f745f268aaeb8cfd5cda312b36b9ede8

Validation

• 226 tests and 18 subtests passed.
• All eleven locale JSON files were validated with 863 matching keys and placeholders.
• Python compilation and configuration JSON validation passed.
• A clean PyInstaller build produced the GUI, EngineHost, MCP server, and embedded Python runtime without optional engine dependencies or local configuration.
• The Inno Setup installer checksum and Windows version metadata were verified.
• An isolated Russian CPU/Piper installation, GUI startup, EngineHost help check, and silent uninstall completed successfully.

LocalText2Voice v1.2.0

What's new

• Added automatic SRT and karaoke-style ASS subtitle sidecars from Faster Whisper word timestamps, including chapter-aware exports and narration offsets for podcast mixes.
• Added optional multilingual text normalization with editable SQLite dictionaries for ten languages and independent rules for numbers, ordinals, dates, currencies, percentages, measurements, and Roman numerals.
• Added a second Whisper quality layer that detects unexplained audio after the last source-aligned word, highlights possible trailing artifacts, and contributes to retry candidate selection.
• Added manual audio-tail cutting and optional conservative automatic cleanup. Auto-cut candidates are transcribed again and kept only when the combined review result improves.
• Persisted review metrics and audiobook output paths so subtitles refresh correctly after verification, regeneration, tail correction, rebuilds, UI jobs, and MCP/local-server jobs.
• Made OmniVoice installation reproducible by resolving the tested OmniVoice and PyTorch versions together, using the correct CPU/CUDA indexes and retrying transient Windows cleanup failures.
• Improved Windows uninstall behavior: users can remove downloaded engines, models, runtime dependencies, and voice-gallery data while keeping projects, exports, settings, music, and logs.
• Updated all ten interface languages, documentation, third-party notices, default settings, and the database/settings schemas.

Windows distribution

• Clean installer for Windows 10/11 x64.
• Piper remains the lightweight default engine.
• Optional TTS engines and Faster Whisper remain on-demand downloads.
• Includes the GUI, persistent EngineHost, MCP stdio server, FFmpeg, Piper, bundled voices/music, and the clean embedded Python runtime.
• The installer is currently unsigned.
• SHA-256: 4605bc7e4d590894c28eb59a897d931baade0d3905b9b01aaa00494ec3d2685b

Validation

• 205 tests and 12 subtests passed.
• All ten locale JSON files were validated with 855 matching keys.
• A clean PyInstaller build produced the GUI, EngineHost, MCP server, and embedded Python runtime without optional engine dependencies or local configuration.
• The installer checksum and Windows version metadata were verified.
• An isolated CPU/Piper installation, GUI startup, and silent uninstall completed successfully.

LocalText2Voice v1.1.2

What's new

• Added General Documentation and Markup Help links to the Help menu.
• Added translated labels for both Help actions in all ten interface languages.
• Added a guided installation dialog for optional TTS engines with requirements, estimated download time, free-space validation, progress, cancellation, and clearer status messages.
• Hardened settings migration and chunk-size validation so incomplete or malformed configurations recover to safe defaults.
• Updated the LTV Markup manual with simpler PLAY examples that demonstrate filename-only and extension-free audio lookup.
• Marked sound-effect and music timeline markup, plus automatic updates, as completed in the roadmap.

Windows distribution

• Clean installer for Windows 10/11 x64.
• Piper remains the lightweight default engine.
• Optional engines and Faster Whisper remain available as on-demand installs.
• The installer is currently unsigned.
• SHA-256: 261ac8854b775dcafafdc43bdc0b81d4f035a2ab6c04899fb2c8244a31779510

Validation

• 166 tests and 12 subtests passed.
• All ten locale JSON files were validated.
• A clean CPU/Piper silent installation was verified from scratch.
• The installed README and LTV Markup manual were checked against this release.

LocalText2Voice v1.1.1

What's new

• Restored the native Windows frame for better compatibility, resizing, maximization, shadows, and Snap Layouts.
• Kept File, Edit, Selection, View, and Help in the standard application menu bar.
• MCP/HTTP generation now returns the editable project path and manifest.
• Added Open Recent for quick access to recent projects.
• Added Open Recent translations for all interface languages.
• Temporarily removed the demo video reference from the README.

Windows distribution

• Clean installer for Windows 10/11 x64.
• Piper included as the lightweight default engine.
• Advanced engines and Faster Whisper remain available as on-demand installs.
• The installer is currently unsigned.
• SHA-256: ed70b091b3fb286f70f3ce3e8810648513a85a42cd2538f517d576aaff4194f2

Validation

• 156 tests and 12 subtests passed.
• A clean CPU/Piper silent installation was verified from scratch.

LocalText2Voice v1.1.0

What's new

• Automatic stable-version checks through GitHub Releases.
• Installer downloads are verified with SHA-256 before execution.
• All bundled background music and effects from music/ are included in the installer.
• Improvements to the advanced audio mix and timeline.
• Clean Windows packaging: Piper and the three intended voices are included, while optional engines and models are downloaded after installation.

Updating from v1.0.0

Users running v1.0.0 need to install v1.1.0 manually once. Stable releases after this version can be detected from Help > Check for updates.

LocalText2Voice v1.0.0

LocalText2Voice v1.0.0

First Windows installer distribution for LocalText2Voice.

Download

Download LocalText2Voice-Setup-1.0.0.exe below and run it on Windows.

What is included

• Windows Inno Setup installer.
• CPU light setup profile with Piper as the default local/offline engine.
• Powerful GPU setup profile that prepares OmniVoice and Faster Whisper on first launch.
• Multi-engine TTS workflow with Piper, Kokoro, Chatterbox, Qwen, OmniVoice, and optional cloud/custom HTTP engines.
• Voice Library with downloadable/reference voices and previews.
• LocalText2Voice markup language for voice/language changes, pauses, speed, volume, and model commands.
• Faster Whisper review workflow for segment verification.
• Audio Mix page for podcast-style music mixing.
• Updated documentation and installer/code-signing notes.

Signing status

This build is currently unsigned. Code signing is planned for a future release once a certificate or open source signing workflow is ready. SHA256 checksum is provided as a separate asset.

SHA256

See LocalText2Voice-Setup-1.0.0.exe.sha256.

LocalText2Voice v0.4.6

LocalText2Voice v0.4.6

Small Chatterbox settings UI fix.

Fixes

• Chatterbox now shows Installed when the current runtime is available and
  usable, even if an older model-cache warmup manifest was left as failed.
• The Remove button is enabled when either the Chatterbox runtime or model
  cache exists, so partial installs can be cleaned up from the UI.

Notes

• Chatterbox CUDA runtime remains distributed separately as
  chatterbox-runtime-v3.

LocalText2Voice v0.4.5

LocalText2Voice v0.4.5

Chatterbox runtime compatibility hotfix.

Fixes

• Fixed Chatterbox Multilingual startup with chatterbox-tts 0.1.7.
• The runtime no longer passes the unsupported t3_model argument to
  ChatterboxMultilingualTTS.from_pretrained().
• The app now targets chatterbox-runtime-v3, so older v2 runtimes are marked as
  Update available and replaced through the normal Install button.

Validation

• Added a unit test that matches the current Chatterbox multilingual loader
  signature.

LocalText2Voice v0.4.4

LocalText2Voice v0.4.4

Small Chatterbox safety hotfix.

Fixes

• Audio generation now refuses to use an outdated Chatterbox runtime and shows a
  clear update message instead.
• Users with an older AppData runtime should open Chatterbox settings and click
  Install once to download the current CUDA runtime pack.

Included from v0.4.3

• Existing Chatterbox runtime installs are checked against the expected runtime
  version.
• Older AppData runtimes are marked as Update available in settings.
• GPU/CUDA detection is available before the Chatterbox runtime download.