voice-mode 2.26.0__tar.gz → 2.28.0__tar.gz

{voice_mode-2.26.0 → voice_mode-2.28.0}/.gitignore

@@ -111,3 +111,9 @@ testdir/
 
 # Profiling output
 *.prof
+
+# Model files (should be downloaded, not committed)
+models/
+*.mlpackage/
+*.mlmodel
+*.mlmodelc/

{voice_mode-2.26.0 → voice_mode-2.28.0}/CHANGELOG.md

@@ -7,6 +7,137 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [2.28.0] - 2025-08-23
+
+### Added
+- **Comprehensive CLI help support**
+  - Added `-h` and `--help` options to all CLI commands and subcommands
+  - Consistent help functionality across all command groups (kokoro, whisper, livekit, config, etc.)
+  - Help options available for both groups and individual commands
+  - Improved user experience with quick access to command documentation
+
+- **Core ML support for whisper.cpp installation**
+  - Whisper install now uses CMake instead of Make for better control
+  - Automatically enables Core ML support on Apple Silicon Macs
+  - Provides ~3x faster encoding performance with Core ML acceleration
+  - Core ML models automatically converted during installation
+  - Falls back gracefully if Core ML conversion fails
+  
+- **Enhanced whisper status command**
+  - Shows whisper.cpp version information
+  - Displays Core ML support status (enabled/disabled)
+  - Shows if Core ML model is active for current model
+  - Reports GPU acceleration type (Metal/CUDA)
+  - Helper utility in `whisper_version.py` for capability detection
+  
+- **Audio conversion optimization for local whisper**
+  - Automatically detects truly local whisper (not SSH-forwarded)
+  - Skips WAV to MP3 conversion for local whisper, sending WAV directly
+  - Adds timing measurements for audio format conversion
+  - Logs conversion time at INFO level for performance monitoring
+  - Significantly reduces STT processing time for local deployments
+  
+- **Whisper model benchmark command**
+  - New `whisper model benchmark` CLI command
+  - Compares performance across multiple models
+  - Shows load time, encode time, and total processing time
+  - Calculates real-time factor for each model
+  - Fixed timing output by removing --no-prints flag
+  - Helps users choose optimal model for speed/accuracy tradeoffs
+  - Provides personalized recommendations based on results
+
+### Fixed
+- **MCP server configuration**
+  - Fixed .mcp.json to use `uv run voicemode` for local development
+  - Removed hardcoded paths for better portability
+  - Works correctly with project-local development version
+  
+- **Whisper model management**
+  - Fixed model active command to properly update configuration
+  - Fixed naming conflict in model install CLI command
+  - Benchmark now correctly shows timing information
+  - Core ML conversion errors are now properly reported and handled
+
+## [2.27.0] - 2025-08-20
+
+### Added
+- **CLI version and update commands**
+  - New `voice-mode version` command to display current version
+  - New `voice-mode update` command to upgrade to latest version
+  - Comprehensive bats tests for version and update functionality
+  - Automatic version detection from package metadata
+  
+- **Shell completion support for CLI**
+  - New `voice-mode completion` command group with bash, zsh, and fish subcommands
+  - Automatic tab completion for all commands, options, and arguments
+  - Install.sh automatically configures shell completions during setup
+  - Native Click completion mechanism for dynamic suggestions
+  
+- **Parallel operations documentation**
+  - Documented `wait_for_response=False` pattern in converse tool
+  - Enables speaking while performing other operations simultaneously
+  - Creates more natural conversations by eliminating dead air
+  - Marked as RECOMMENDED pattern with clear usage examples
+  
+- **Comprehensive Whisper model management system**
+  - New `whisper models` CLI command to list all available models with status
+  - `whisper model active` command to get/set the active model  
+  - `whisper model install` and `whisper model remove` commands
+  - Model registry with complete size/hash metadata for all Whisper models
+  - Color-coded output showing installed/available models (green=installed, yellow=selected)
+  - Support for English-only models and all multilingual variants
+  - Automatic Core ML conversion on Apple Silicon for improved performance
+  - Shell completion support for all model management commands
+  
+- **MCP tools for model management**
+  - `list_models` tool to list all available Whisper models with status
+  - Enhanced `download_model` tool with registry validation
+  - Force download option to re-download corrupted models
+  - Skip Core ML option for testing
+  - Parity between CLI and MCP interfaces
+  
+- **Infrastructure improvements**
+  - Centralized model registry in `whisper/models.py` with all model metadata
+  - Model categorization: tiny, base, small, medium, large, turbo
+  - Size information for all models (39MB to 3.1GB)
+  - SHA256 hashes for integrity verification
+  - Shared download logic extracted to helpers module
+  - Dynamic Click-based shell completions replacing static files
+  - Comprehensive test suite for model management
+
+### Changed
+- **Configuration file naming**
+  - Renamed `.voicemode.env` to `voicemode.env` (removed leading dot)
+  - Added backwards compatibility to check for old filename
+  - Shows deprecation warning when old filename is used
+  - Updated all documentation to reference new filename
+  - Updated systemd service templates
+  
+- Replaced static shell completions with Click-generated dynamic completions
+- Shell completion files now generated from CLI structure
+- Whisper model downloads now use centralized registry for validation
+- Model status checks now verify both file existence and selection
+
+### Fixed
+- **macOS installation improvements**
+  - Added coreutils dependency for timeout command support
+  - Fixed duplicate launchctl load in service installers
+  - Improved zsh PATH configuration by sourcing profile after UV/npm additions
+  - Skip sudo prompts on macOS to prevent installation issues
+  
+- **Test suite fixes**
+  - Fixed deprecation warning appearing in help output
+  - Renamed deprecated `.voicemode.env` to `voicemode.env` to fix test failures
+  
+- Whisper model management now properly uses voicemode.env configuration file
+- Test suite updated for all API changes and return value structures
+- Resolved all CI test failures related to service status and diagnostics
+
+### Removed
+- Old static shell completion files  
+- SERVICE_COMMANDS.md (replaced by integrated CLI commands)
+- Shell aliases file (functionality moved to Click commands)
+
 ## [2.26.0] - 2025-08-18
 
 ### Added

{voice_mode-2.26.0 → voice_mode-2.28.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: voice-mode
-Version: 2.26.0
+Version: 2.28.0
 Summary: VoiceMode - Voice interaction capabilities for AI assistants (formerly voice-mcp)
 Project-URL: Homepage, https://github.com/mbailey/voicemode
 Project-URL: Repository, https://github.com/mbailey/voicemode
@@ -98,6 +98,10 @@ Natural voice conversations for AI assistants. Voice Mode brings human-like voic
 1. **🎤 Computer with microphone and speakers** OR **☁️ LiveKit server** ([LiveKit Cloud](https://docs.livekit.io/home/cloud/) or [self-hosted](https://github.com/livekit/livekit))
 2. **🔑 OpenAI API Key** (optional) - Voice Mode can install free, open-source transcription and text-to-speech services locally
 
+**Optional for enhanced performance:**
+
+- **🍎 Xcode** (macOS only) - Required for Core ML acceleration of Whisper models (2-3x faster inference). Install from [Mac App Store](https://apps.apple.com/app/xcode/id497799835) then run `sudo xcode-select -s /Applications/Xcode.app/Contents/Developer`
+
 ## Quick Start
 
 > 📖 **Using a different tool?** See our [Integration Guides](docs/integrations/README.md) for Cursor, VS Code, Gemini CLI, and more!

{voice_mode-2.26.0 → voice_mode-2.28.0}/README.md

@@ -29,6 +29,10 @@ Natural voice conversations for AI assistants. Voice Mode brings human-like voic
 1. **🎤 Computer with microphone and speakers** OR **☁️ LiveKit server** ([LiveKit Cloud](https://docs.livekit.io/home/cloud/) or [self-hosted](https://github.com/livekit/livekit))
 2. **🔑 OpenAI API Key** (optional) - Voice Mode can install free, open-source transcription and text-to-speech services locally
 
+**Optional for enhanced performance:**
+
+- **🍎 Xcode** (macOS only) - Required for Core ML acceleration of Whisper models (2-3x faster inference). Install from [Mac App Store](https://apps.apple.com/app/xcode/id497799835) then run `sudo xcode-select -s /Applications/Xcode.app/Contents/Developer`
+
 ## Quick Start
 
 > 📖 **Using a different tool?** See our [Integration Guides](docs/integrations/README.md) for Cursor, VS Code, Gemini CLI, and more!

{voice_mode-2.26.0 → voice_mode-2.28.0}/voice_mode/__version__.py

@@ -1,3 +1,3 @@
 # This file is automatically updated by 'make release'
 # Do not edit manually
-__version__ = "2.26.0"
+__version__ = "2.28.0"