Skip to content

Releases: sl5net/SL5-aura-service

Release v0.31.01

12 Mar 06:03

Choose a tag to compare

Release v0.31.01

🚀 Highlights

  • Massive Performance Gain: CLI request latency reduced from ~14s to ~340ms.
  • Advanced TTS Workflow: Piper server integration with automatic espeak-ng fallback.
  • Peter AI Agent: Added Peter AI agent v0.6.7 including enhanced search tools.

🧠 Core & Docs

  • Internationalization: Massive translation update (170+ files) and reorganization into .i18n subdirectories.
  • Resource Optimization: Dynamic LanguageTool JVM configuration based on system resources.
  • Stability: Fixed input field visibility issues and improved LanguageTool server readiness checks.
  • Documentation: Added GettingStarted.md, video sections, and new demo visualisations.

Ultra-Fast. Intelligent. Multilingual. Aura.

ko-fi

Full Changelog: v0.30.00...v0.31.00

Release v0.31.00

10 Mar 09:48

Choose a tag to compare

Release v0.31.00

🚀 Highlights

  • Massive Performance Gain: CLI request latency reduced from ~14s to ~340ms.
  • Advanced TTS Workflow: Piper server integration with automatic espeak-ng fallback.
  • Peter AI Agent: Added Peter AI agent v0.6.7 including enhanced search tools.

🧠 Core & Docs

  • Internationalization: Massive translation update (170+ files) and reorganization into .i18n subdirectories.
  • Resource Optimization: Dynamic LanguageTool JVM configuration based on system resources.
  • Stability: Fixed input field visibility issues and improved LanguageTool server readiness checks.
  • Documentation: Added GettingStarted.md, video sections, and new demo visualisations.

Ultra-Fast. Intelligent. Multilingual. Aura.

ko-fi

Full Changelog: v0.30.00...v0.31.00

Release v0.30.00

06 Mar 04:51

Choose a tag to compare

🚀 Highlights

  • Auto-Maintenance: Radio-Aura Cache-Generierung (4s Stille-Timer).
  • X11-Stabilität: "Maximum clients"-Leak timeout xdotool fixed.

🧠 Core & Web

  • Streamlit TTS: Speak suport Web & Android .
  • Radio v1.3.2: Tracking & Deep-Dive updated.

Layout-Aware. Native. Unstoppable. Aura.

ko-fi

Full Changelog: v0.29.00...v0.30.00

Release v0.29.00

02 Mar 08:54

Choose a tag to compare

Release v0.29.00

🚀 Highlights

This release introduces significant improvements to the Text-to-Speech (TTS) engine, adds the Radio Deep-Dive generator, and implements advanced self-healing for configuration files. We’ve also enhanced multilingual support for dotool and added automated error reporting via voice commands.


🎙️ Speech & Audio Enhancements

  • Dual-Voice Speech: Added support for parallel speech using both Piper (high-quality neural TTS) and espeak-ng.
  • Speech Lock: Implemented _speech_lock to prevent Piper and espeak from overlapping.

🧠 Core Engine & Self-Healing

  • Map Health Check: New time-sensitive health checks for mapping files. The system now automatically detects and heals swapped fields (ID vs. Regex) in map_reloader.py.
  • Misrecognition Logging: Voice-triggered error reporting. Say "fehler melden" to automatically extract the last STT error from logs and save it to docs/bugfix/TODO/misrecognitions.md.
  • Streamlined Handshake: Improved Python-to-Bash communication via temp files, allowing the shell script to read input methods and delays immediately upon heartbeat.
  • Performance: Implemented dynamic log filtering and optimized regex caching.

⌨️ Input & Automation

  • dotool Evolution:
    • Implemented FIFO support for high-speed input.
    • Dynamic Language Detection: Automatically detects keyboard layout from config/model_name.txt.
    • Layout Fixes: Resolved Y/Z swaps and missing German Umlauts (ä, ö, ü, ß) in dotool.
  • Exclusion Rules: Added exclude_windows parameter to rules to prevent triggers in specific applications.

📻 New Tools & Plugins

  • Radio Deep-Dive (v1.3.2): A new generator with file tracking and GitHub link generation for deep-dive audio content.
  • Ollama Integration: Added ollama_speak_answers_only_espeak_loop.py for specialized LLM interaction loops.
  • Plugin Cache: Refactored structure; moved plugin_call_cache.sqlite to a dedicated data/ folder.

🛠 Tech Debt & Fixes

  • Improved error handling in the Translation plugin.
  • Refactored settings_local_log_filter to config.filters.
  • Fixed type signature in setup_project_structure.
  • Added DEV_MODE_show_window_title_stuff for easier debugging of window-matching rules.

Layout-Aware. Native. Unstoppable. Aura.

ko-fi

Full Changelog: v0.28.00...v0.29.00

v0.28.00 - The "Rock-Solid Wayland Input" Update

27 Feb 19:03
302e8ec

Choose a tag to compare

v0.28.00 - The "Rock-Solid Wayland Input" Update

⚠️ Important for Linux (Wayland) Users:
This version introduces dotool as the primary engine for Wayland keyboard automation. To ensure a smooth experience, please install dotool and add your user to the input group:

sudo pacman -S dotool

or maybe

sudo yay -S dotool

sudo usermod -aG input $USER (Then log out and back in)

🚀 Highlight: Aura now features Native Wayland Keyboard Emulation. By migrating from xdotool to dotool, we have finally solved the notorious "keyboard sticking" issue under Wayland.

Overview
Aura v0.28.00 focuses on input reliability and linguistic precision. We’ve overhauled the way text is injected into the system, making it faster, safer, and aware of your local keyboard layout. Whether you are typing complex URLs or German Umlaute, Aura now handles them natively on Wayland.


System Requirements & Compatibility

  • Wayland Input Engine:New: dotool Integration (Bypasses XWayland for superior stability)
  • Keyboard Layouts:Full Support for QWERTZ/German (Fixes Z/Y swap and special characters like ://)
  • Desktop Environments:Plasma 6 / CachyOS (Verified stability)

What’s New in v0.28.00

⌨️ Native Wayland Input (dotool)

  • Layout-Aware Automation: Migrated to dotool for Wayland sessions. This ensures that special characters (ö, ä, ü, ß) and symbols (:, /) are typed correctly according to your system's keyboard layout, eliminating the "US-layout fallback" bug.
  • Modifier Recovery (Anti-Sticking): Implemented an advanced cleanup trap that explicitly releases Shift, Alt, and Control keys on script exit. This prevents the "ghost-keys" effect where keys appear to stay pressed after a dictation.
  • Safe String Injection: Switched from echo to printf for all keyboard buffers to safely handle complex strings, backslashes, and leading dashes without triggering shell-command errors.

🛡️ Resilience & Robustness

  • Hybrid Input Logic: Aura now dynamically chooses the best tool for the job. It uses native dotool for modern Wayland environments while maintaining xdotool with optimized delays for legacy X11 sessions.
  • Improved Transcription Cleaning: Refined the sanitize_transcription_start function to remove invisible Unicode characters (ZWNBSP/ZWSP) that often caused silent failures in text processing.

⚙️ Automation & Macros

  • 0 A.D. Game Macros: Optimized the performance of game-specific macros. While text input is now native, high-level window actions still utilize the XWayland bridge where necessary to maintain compatibility with legacy game engines.
  • Auto-Enter Logic: Fixed the Auto-Return feature for Wayland, ensuring that the "Submit" action works seamlessly across all window types.

Upgrade & Migration

  1. Run git pull to upgrade to v0.28.00.
  2. Install dotool: On Arch-based systems (CachyOS/Manjaro), run sudo pacman -S dotool or use yay or someting ...
  3. Permissions: Ensure your user is in the input group: sudo usermod -aG input $USER. A logout/login is required for this to take effect.
  4. X11 Users: No changes required; the script will automatically fall back to your existing xdotool setup.

Layout-Aware. Native. Unstoppable. Aura.

ko-fi

Full Changelog: v0.26.03...v0.28.00

v0.27.01 - The Wayland & Context-Aware Automation Update

27 Feb 01:12

Choose a tag to compare

⚠️ Important for Linux Users:
After upgrading, please run the following command to make the new key-management script executable:
chmod +x tools/keep-keys-up.sh

v0.27.01 - The Wayland & Context-Aware Automation Update

🚀 Highlight: Aura now officially supports KDE Plasma 6 (Wayland). By leveraging the AT-SPI (Accessibility) Bus, we’ve bypassed the strict security isolation of Wayland to enable context-aware window tracking for a seamless STT experience.

Overview
Aura v0.27.01 is a major step toward modern Linux desktop compatibility and advanced local automation. We’ve introduced voice-activation, a robust notification system, and deep integration with local waste disposal schedules (Wannweil plugin suite).


System Requirements & Compatibility

  • Linux (Plasma 6 / Wayland):New: Full Support (Via AT-SPI Accessibility Bus)
  • Nvidia / CachyOS:Verified Compatibility (Tested on Python 3.14 environments)
  • Email Notifications:Gmail SMTP Integration (App Password support)

What’s New in v0.27.01

🖥️ Plasma 6 & Wayland Integration

  • AT-SPI Window Tracking: Implemented a sophisticated retrieval system for active window titles on Wayland. By acting as an accessibility client, Aura can now "see" window focus where traditional D-Bus methods fail.
  • Audio & TTS Diagnostics: Added AUDIO_DIAGNOSTICS_EN.md and espeak_check.py to ensure reliable speech output across different Linux distributions.
  • Keyboard Management: Integrated setxkbmap options to manage Caps Lock behavior, ensuring a consistent dictation environment.

🤖 Voice Activation & Interaction

  • Voice Wake-Word: Introduced a new voice-activation rule. You can now toggle the engine’s active state entirely by voice, reducing the need for manual hotkeys.
  • Improved Dictation Scripts: Updated to start_dictation_v2.1.bat for better path handling and startup reliability.

📅 The "Wannweil" Automation Suite

  • Automated Waste Disposal Reminders: A new robust plugin that parses 12-month silo-based PDF schedules and provides automated reminders.
  • Gmail SMTP Notifications: Integrated secure email alerts via Gmail App Passwords. Includes weekday localization and human-readable date formatting.
  • Clean Audio Alerts: Enhanced espeak reliability using synchronous subprocess calls and automated emoji-stripping for clearer voice notifications.

🛠️ Infrastructure & CI Performance

  • Optimized GitHub CI: Excluded large language models (0.21, 0.22) from CI runs to significantly reduce build times and resource consumption.
  • Path Consolidation: Massive refactor of the project root and subfolder structure to ensure robust cross-platform path synchronization.
  • Ghost Plugin System: Added a deactivation hint system for "Ghost Plugins" to keep the engine clean and modular.

🛡️ Fixes & Refinements

  • Timing Accuracy: Fine-tuned time.sleep intervals (0.05s) in critical test pipelines for better synchronization during high-load scenarios.
  • Linguistic Cleanup: Fixed several typos in core processing rules (e.g., \w*bersetzung) and improved wildcard handling in the check_trash utility.
  • Pathlib Migration: Strategically replaced some pathlib.Path calls with more robust string-based path handling to prevent OS-specific parsing errors.

Upgrade & Migration

  1. Run git pull to upgrade to v0.27.01.
  2. Linux Users: If using Wayland, ensure you follow the new setup in docs/ to enable the Accessibility Bus (QT_ACCESSIBILITY=1).
  3. Re-run pip install -r requirements.txt if you have recently updated your system Python version (e.g., to 3.14).

Context-Aware. Wayland-Ready. Aura.

ko-fi

v0.26.03 - The Performance & Multi-Core Optimization Update

31 Jan 20:05

Choose a tag to compare

v0.26.03 - The Performance & Multi-Core Optimization Update

🚀 Highlight: This release introduces a massive boost in processing speed. By implementing full multi-core parallelization, we reduced the core logic self-test duration from 60 seconds to just 11 seconds.

Overview
Aura v0.26.03 focuses on extreme performance and CI resilience. We have refactored the engine's core validation and LanguageTool integration to scale dynamically with modern high-end CPUs.


System Requirements & Compatibility

  • Linux (Ubuntu/Manjaro):Enhanced Support (Resolved CI package 404 conflicts)
  • Multi-Core CPUs:Fully Optimized (Dynamic scaling across all available processor threads)

What’s New in v0.26.03

⚡ Extreme Performance & Parallelism

  • Process-Level Parallelization: Replaced sequential testing with a high-performance ProcessPoolExecutor. The engine now bypasses the Python GIL to utilize all available CPU cores, resulting in a 4x speed increase during validation.
  • PID-Aware Lazy Sessions: Optimized LanguageTool API calls by implementing a persistent session system that is isolated per Process ID (PID). This ensures 100% stability and connection reuse even under heavy parallel load.
  • Race Condition Resolution: Fixed a critical bug in process_text_in_background.py where redundant timestamps caused file collisions. Tasks now use isolated directories and unique indices for flawless concurrent processing.

🛡️ CI Resilience & Infrastructure

  • Ubuntu Noble CI Fix: Added apt-get update -y to the GitHub Actions workflow to resolve 404 Not Found errors during setup, ensuring reliable builds on Ubuntu 24.04.
  • Granular Throttling: Refined the run_always_ignore_times logic to prevent unnecessary task skips while maintaining system responsiveness.

🏗️ Refactoring & Modernization

  • Context-Aware Signatures: Refactored the signature system to use global state management, making text-cleaning more reliable and aware of the current processing context.
  • Linguistic Logic Restoration: Re-enabled key conditions in the text processing pipeline to ensure full linguistic accuracy across all supported languages.

📖 Legal & Docs

  • German License Context: Added explain_in_german_deutsch_License.md to improve legal transparency for the German-speaking developer community.
  • Readme & Links: Updated documentation links and internal references for better navigation.

Upgrade & Migration

  1. Run update.bat or git pull to upgrade to v0.26.03.
  2. The engine will automatically utilize your CPU's full core count for the next self-test.
  3. Enjoy the significantly reduced wait times!

Speed. Security. Stability. Aura.

ko-fi

v0.26.02 - The Windows Stability & Automation Update

30 Jan 08:55

Choose a tag to compare

v0.26.02 - The Windows Stability & Automation Update

🚀 Highlight: This release introduces high-privilege automation via Windows Task Scheduler and a robust "State-Machine" for file-watching, effectively solving system-key (F10/F11) interception and double-typing issues.

Overview
Aura is an offline engine that turns voice into actions and text across platforms. This version focuses on bringing Windows reliability on par with Linux by overcoming OS-level file locking and privilege restrictions.


System Requirements & Compatibility

  • Windows:Enhanced Support (Now uses Task Scheduler for high-privilege automation)
  • macOS / Linux: ✅ Linux (Wayland): ⚠️ Limited / Experimental.

What’s New in v0.26.01

🛠 Windows Reliability & Privilege Automation

  • High-Privilege Hotkeys: Aura now automates the creation of a Windows Task Scheduler entry during setup. This allows trigger-hotkeys.ahk to run with "Highest Privileges" on logon, enabling reliable interception of system keys like F10/F11 even in elevated applications—all without manual UAC prompts.
  • Improved Power Management: Added DontStopIfGoingOnBatteries and other critical flags to background tasks to ensure Aura keeps running on laptops during battery use.
  • Silent Background Lifecycle: Implemented a Heartbeat & Self-Close mechanism combined with #SingleInstance Off. This allows seamless, silent updates of AHK scripts without "Could not close previous instance" dialogs.
  • PowerShell-Driven Cleanup: Improved installer and updater logic using robust PowerShell process management to ensure a clean state before restarts.

✍️ Intelligent Dictation & "Zombie" Protection

  • Double-Typing Fix: Introduced a State-Machine Map in type_watcher.ahk. It tracks file states (Queued/Processing/Done) and implements a "memory" to ignore redundant Windows file-system events, ensuring every dictated word is typed exactly once.
  • Robust File Deletion: Added a retry-logic for the "Zombie" file cleanup to handle Windows-specific file access locks during active TTS output.
  • Editor Sync: Added sync_editor.ahk to force saves and auto-confirm "Reload File" dialogs in editors like Notepad++, providing a fluid real-time dictation experience.

🔍 Engine Intelligence & Security

  • Privacy Taint Tracking: Implemented a new mechanism to track and protect sensitive data flow within the engine.
  • Catch-all Regex Warnings: The engine now warns developers if a rule (like .+) is blocking critical components like LanguageTool.
  • Bytecode Protection: Disabled Python bytecode generation for cleaner project roots and easier auditing.

📦 Tooling & Installer Enhancements

  • Professional Toolset: Installer now optionally includes Notepad++, OpenJDK 17, Double Commander, and Glogg (for high-performance log reading).
  • Automated Permissions: Added fix_permissions.bat and SystemCheck.ps1 to resolve common environment issues automatically.
  • Developer DX: Added map_tagger.py improvements, lowercase map normalization for cross-platform consistency, and detailed error descriptions in self-tests.

🌐 Internationalization & Support

  • Support Infrastructure: Added a dedicated Contact & Support page and comprehensive documentation for AHK internals (docs/windows/AHK_SCRIPTS.md).
  • Multilingual Foundation: Expanded README and documentation to support all project languages.

Upgrade & Migration

  1. Run update.bat to pull the latest changes.
  2. The setup script will attempt to register the new Task Scheduler entries (Admin rights required once during setup).
  3. If you previously had issues with F10/F11 on Windows, they should now be resolved automatically.

Speed. Security. Stability. Aura.

ko-fi

v0.26.01 - The Windows Stability & Automation Update

29 Jan 14:33

Choose a tag to compare

v0.26.01 - The Windows Stability & Automation Update

🚀 Highlight: This release introduces high-privilege automation via Windows Task Scheduler and a robust "State-Machine" for file-watching, effectively solving system-key (F10/F11) interception and double-typing issues.

Overview
Aura is an offline engine that turns voice into actions and text across platforms. This version focuses on bringing Windows reliability on par with Linux by overcoming OS-level file locking and privilege restrictions.


System Requirements & Compatibility

  • Windows:Enhanced Support (Now uses Task Scheduler for high-privilege automation)
  • macOS / Linux: ✅ Linux (Wayland): ⚠️ Limited / Experimental.

What’s New in v0.26.01

🛠 Windows Reliability & Privilege Automation

  • High-Privilege Hotkeys: Aura now automates the creation of a Windows Task Scheduler entry during setup. This allows trigger-hotkeys.ahk to run with "Highest Privileges" on logon, enabling reliable interception of system keys like F10/F11 even in elevated applications—all without manual UAC prompts.
  • Improved Power Management: Added DontStopIfGoingOnBatteries and other critical flags to background tasks to ensure Aura keeps running on laptops during battery use.
  • Silent Background Lifecycle: Implemented a Heartbeat & Self-Close mechanism combined with #SingleInstance Off. This allows seamless, silent updates of AHK scripts without "Could not close previous instance" dialogs.
  • PowerShell-Driven Cleanup: Improved installer and updater logic using robust PowerShell process management to ensure a clean state before restarts.

✍️ Intelligent Dictation & "Zombie" Protection

  • Double-Typing Fix: Introduced a State-Machine Map in type_watcher.ahk. It tracks file states (Queued/Processing/Done) and implements a "memory" to ignore redundant Windows file-system events, ensuring every dictated word is typed exactly once.
  • Robust File Deletion: Added a retry-logic for the "Zombie" file cleanup to handle Windows-specific file access locks during active TTS output.
  • Editor Sync: Added sync_editor.ahk to force saves and auto-confirm "Reload File" dialogs in editors like Notepad++, providing a fluid real-time dictation experience.

🔍 Engine Intelligence & Security

  • Privacy Taint Tracking: Implemented a new mechanism to track and protect sensitive data flow within the engine.
  • Catch-all Regex Warnings: The engine now warns developers if a rule (like .+) is blocking critical components like LanguageTool.
  • Bytecode Protection: Disabled Python bytecode generation for cleaner project roots and easier auditing.

📦 Tooling & Installer Enhancements

  • Professional Toolset: Installer now optionally includes Notepad++, OpenJDK 17, Double Commander, and Glogg (for high-performance log reading).
  • Automated Permissions: Added fix_permissions.bat and SystemCheck.ps1 to resolve common environment issues automatically.
  • Developer DX: Added map_tagger.py improvements, lowercase map normalization for cross-platform consistency, and detailed error descriptions in self-tests.

🌐 Internationalization & Support

  • Support Infrastructure: Added a dedicated Contact & Support page and comprehensive documentation for AHK internals (docs/windows/AHK_SCRIPTS.md).
  • Multilingual Foundation: Expanded README and documentation to support all project languages.

Upgrade & Migration

  1. Run update.bat to pull the latest changes.
  2. The setup script will attempt to register the new Task Scheduler entries (Admin rights required once during setup).
  3. If you previously had issues with F10/F11 on Windows, they should now be resolved automatically.

Speed. Security. Stability. Aura.

ko-fi

v0.24.02 - Path Length & Syntax Hotfixes

16 Jan 15:55

Choose a tag to compare

v0.24.02 - Path Length & Syntax Hotfixes

🚀 Hotfix: Added critical startup check for Windows MAX_PATH limits and fixed syntax errors(not critical) in the Fuzzy Path Navigator plugin.

Overview
System-Wide Offline Voice to Commands or Text, Pluggable System. Aura is an offline engine that turns voice into actions and text across platforms.


System Requirements & Compatibility

  • Windows: ✅ Fully supported (AutoHotkey / PowerShell)
  • macOS: ✅ Fully supported (AppleScript)
  • Linux (X11/Xorg): ✅ Fully supported (recommended)
  • Linux (Wayland): ⚠️ Limited / Experimental — global hotkeys & auto-typing often unavailable

**Highlights **

  • Improved offline LLM integrations and performance:
    • Offline LLM fallback (Ollama / Llama 3.2) and Instant Mode for low-latency replies.
    • RAG-light for project documentation support and intelligent SQLite caching with auto-warming.
  • Robustness & reliability:
    • Keep-alive watchdog and long-term memory leak analyzer.
    • Adaptive memory/model prioritization to prevent unloading of active models.
    • Self-healing startup (robust directory creation) and improved watcher implementations for Windows/macOS/Linux.
  • Developer experience & CI:
    • Pre-commit hook, startup config validation, and static analysis checks integrated.
    • Mock-testing suite and several mock tests added (dictation / core logic).
  • Plugins & extensibility:
    • Hot-reloading for map files and on_reload() lifecycle hook for plugins.
    • Enhanced plugin API (language-aware TTS, auto-enter behavior) and new plugins collected.
  • Speech & dictation:
    • Dynamic VAD for responsive dictation and mixed mic+desktop audio capture on Linux.
    • Better dictation session model and advanced automatic text replacement (multi-pass).
  • Tools & packaging:
    • pyzipper for secure zipping, zip_me_nopassword for non-sensitive assets.
    • Installer improvements: cache-first setup logic and package filtering (exclude arg).
  • UX & internationalization:
    • Added German start triggers, Japanese live-translation support, Arabic enablement, and language exclusion options in setup scripts.
  • Miscellaneous notable improvements:
    • Flashcard/Anki quiz progress tracking and MC voice-controlled quiz enhancements.
    • Fuzzy matching, rule normalization, pre-correction FUZZY_MAP, and many tooling improvements (map_tagger, search_rules.sh).
    • CopyQ exporter enhancements, folder-based tags, and auto-typing/export reliability improvements.

Cumulative features note
This release continues building on all previous features (≈110 recorded feats). For a complete, searchable index of implemented features, see the cumulative feature index (maintain a single canonical list in repo/docs). Use per-release pages for highlights and link to the cumulative index for full history.


Upgrade & Migration

  • Backup current config and data before upgrading.
  • If you use Wayland and rely on global hotkeys or auto-typing, switch to an X11 session for full functionality.
  • After upgrade: run the integrated self-test suite and integrity checks (pre-commit/startup validations).

How to help / Contribution pointers

  • Tag PRs with conventional prefixes (feat:, fix:, breaking:) and include PR/issue links to be auto-indexed.
  • Add short metadata (intro release, component, short description) to feature entries when merging to keep the cumulative index accurate.
  • Run the mock tests and CI locally; add test coverage for new features.

Speed. Security. Stability. Aura.

ko-fi