Release Notes - v0.5.0

New Features

Advanced Tagging System

• Multi-tag support: Assign multiple tags to recordings with priority ordering
• Custom prompts per tag: Each tag can have its own summarization prompt that automatically applies
• ASR defaults per tag: Configure default language and speaker settings for each tag
• Tag-based search: Filter recordings by tags using the search bar (e.g., tag:meeting)
• Visual tag management: Beautiful tag display with numbered priorities and easy removal

Enhanced ASR (Automatic Speech Recognition) Integration

• Advanced options UI: Configure language and speaker detection directly from upload screen
• Automatic speaker detection: No more hardcoded speaker limits - let ASR auto-detect participants
• Improved diarization: Better speaker identification with automatic participant extraction
• Simplified configuration: Just set USE_ASR_ENDPOINT=true and the system handles the rest

Document Export

• Word document generation: Export summaries and notes as properly formatted .docx files
• Server-side generation: Reliable document creation using python-docx
• Rich formatting: Headers, bold text, lists, and metadata preserved in exports

Improvements

User Interface

• Improved styling: Better theme consistency for chat bubbles and recording cards
• Reorganized layouts: Recording cards now show title → date → participants → tags
• Icon enhancements: Added visual icons for dates and participants
• Compact design: More efficient use of screen space with appropriate spacing
• Upload progress management: Clear completed uploads with one click

Audio Processing

• Configurable chunk sizes: Respects CHUNK_SIZE_MB environment variable properly
• Better chunk calculation: Improved algorithm reaches target sizes more accurately (95% accuracy)
• AAC file handling: Special handling for problematic AAC-encoded files
• MIME type detection: Enhanced support for M4A, AAC, and other audio formats

Summary Generation

• Better markdown formatting: Improved prompts for consistent bold key-value pairs
• Multiple prompt sources: Combines tag prompts with user prompts in priority order
• Debugging support: Added logging for troubleshooting empty summaries

Bug Fixes

• Fixed NameError when using local LLM setups without API keys
• Fixed UnboundLocalError when uploading files without tags
• Fixed ASR parameter hierarchy (user input > tag defaults > env vars > auto-detect)
• Fixed database migration issues with 'order' column quoting
• Fixed 'No file provided' error on page refresh
• Fixed duplicate tag display in management interfaces
• Fixed full-screen drop zone conflicts causing double uploads
• Fixed code block styling for proper dark/light mode contrast

Configuration

New Environment Variables

• LOG_LEVEL: Control logging verbosity (ERROR/INFO/DEBUG)
• CHUNK_SIZE_MB: Configure audio chunk size for processing (default: 25MB)

Simplified ASR Setup

• Automatic diarization when ASR endpoint is enabled
• Removed redundant configuration options
• Cleaner environment file examples

Documentation

• Improved environment file examples with clear comments
• Added debugging guidance for summary generation issues

Breaking Changes

• None - all changes are backward compatible

Migration Notes

• Database will automatically migrate to support new tagging system
• Existing recordings will work without tags
• ASR users should review simplified configuration in env.asr.example

Coming Next

• Bulk tag operations
• Tag templates for common workflows