Changelog — 1.x

Improved Reliability

• •Automatic task recovery - Long-running tasks now survive network interruptions
• •Voice synthesis jobs can now be recovered if your connection drops mid-generation
• •Progress is preserved - return to the page and continue where you left off
• •Real task cancellation - cancel button now stops background processing

User Experience

• •Unified progress indicators across all admin features
• •Status badges show task progress in list views
• •Automatic reconnection when returning to pages with active tasks

New Admin Features

• •Network Administration - New dashboard for managing infrastructure connections
• •Manage server peers with platform-specific setup guides
• •Configure hostname routing for different services
• •Monitor custom services alongside built-in system status

Dashboard Improvements

• •Consolidated admin dashboard with cleaner layout
• •Consistent color theme across infrastructure management pages
• •Improved infrastructure monitoring in System Status

Performance Improvements

• •10-20x faster translation - Files that took 2-4 hours now complete in 5-10 minutes
• •Batch processing translates segments in groups of 50 instead of one-by-one
• •Same translation quality with identical timestamp synchronization

Enhanced Progress Feedback

• •Real-time updates stream instantly as each batch completes
• •Clear time estimates showing elapsed time and remaining time
• •Batch visibility displays progress like "Batch 4/8 complete"

Improved Reliability

• •Automatic fallback if primary AI service has issues
• •Retry logic handles transient failures automatically
• •Comprehensive logging for troubleshooting

Performance Improvements

• •Consolidated voice synthesis utilities for better efficiency
• •Reduced code duplication across audio processing modules
• •Streamlined date and time formatting across the platform

Codebase Cleanup

• •Removed deprecated and unused code paths
• •Cleaned up legacy test utilities
• •Removed scaffolded features that were never completed

Documentation Updates

• •Archived outdated technical documentation
• •Consolidated deployment guides into single authoritative documents
• •Updated documentation index with current feature references

ScribeFlow Translation

• Fixed issue where editing or downloading a translation would return the original transcription
• Improved language detection for translated content
• Fixed compatibility issue with certain AI translation models

ScribeFlow Article Generation

• Articles now properly track their output language
• Generate articles in multiple languages for the same project

Image Upload Fixes

• Fixed author avatar images not displaying after upload
• Resolved 404 errors when viewing newly uploaded profile pictures
• Fixed author avatars on article pages

Voice Selection

• Fixed voice dropdown not loading in author management
• Voice options now display friendly names

AI-Powered Islamic Rulings Search

• •Search Islamic rulings from shia-islam.org using natural language
• •Semantic search understands meaning, not just keywords
• •Autocomplete suggestions as you type
• •Islamic term synonyms (wudu/ablution, salat/prayer, ghusl/ritual bath)

Study Interface

• •Conversational Q&A with follow-up questions
• •AI provides answers with mandatory source citations
• •Links to original rulings on shia-islam.org
• •Thinking indicator shows AI reasoning process

Supported Marjas

• •Ayatullah Sistani
• •Imam Khamenei

Coming Soon: Ayatullah Makarem Shirazi, Ayatullah Jawadi-Amoli, Ayatullah Araki

Smart Audio Content Detection

• •Audio generation now detects if page content has changed since last generation
• •Skips unnecessary regeneration when content is unchanged (saves time and resources)
• •Option to force regenerate if needed
• •Improved content hash accuracy for reliable change detection

Infrastructure Flexibility

• •System Status page now supports configurable service endpoints
• •Infrastructure changes no longer require code modifications
• •Per-environment configuration for development and production
• •Audio processing services consolidated for improved performance

Transcription Progress

• •Fixed progress stuck at "Starting AI Transcription"
• •Progress events now stream in real-time during transcription
• •Added connection keepalive to prevent timeout during long operations
• •Fixed job recovery showing "processing" after completion

Enhanced Progress Display

• •Step-based progress shows "Step X of Y: [Step Label]"
• •Elapsed time display (mm:ss) during transcription
• •5-second heartbeats keep UI responsive during long operations
• •Dynamic step counting based on pipeline configuration

AI Transcription Settings

• •Fixed model selection validation (removed legacy requirement)
• •Corrected AI model references
• •Settings page now fetches available models dynamically

Publication Theme & Layout Tab

• •New 4th tab in Publication admin for quick theme customization
• •Color pickers for primary, accent, background, text, heading, and link colors
• •Typography selection with curated headline and body font options
• •Homepage style selector: Feature, Media Feature, or Newspaper layouts
• •Live preview of theme changes before saving

Custom Domain Management

• •Automatic DNS configuration for publication domains
• •One-click domain setup with DNS propagation tracking
• •Visit button appears when domain is fully configured
• •Support for Substack-style URL patterns (/p/, /t/, /about, /archive)

AI-Powered Transcription

• •Cloud-based transcription using advanced AI models
• •Automatic audio compression for optimal processing
• •Smart file chunking for long audio files
• •Automatic handling of processing limits

Enhanced Voice Generation

• •Hardware-accelerated voice synthesis for faster generation
• •54+ natural-sounding voices available
• •Up to 30x faster audio generation than previous version
• •Voice cloning with natural expressions (laughs, sighs, pauses)

Infrastructure Improvements

• •Simplified backend architecture for better reliability
• •Reduced codebase complexity by 94%
• •Freed server resources for future features

Click-to-Play Adhaan

• •When you click a prayer notification, the adhaan will now play automatically
• •Works even when the browser was closed (notification click counts as user interaction)
• •Uses your selected sound preference (Shia Adhaan or Sunni Adhaan)
• •If you have Beep or Silent selected, no audio will play on click

ScribeFlow UI Simplification

• •Removed Quality Tier selector - fixed pipeline means automatic optimization
• •Simplified Content Type descriptions for clearer guidance
• •AI Transcription info now always visible for transparency

Prayer Time Notifications

• •Fixed Maghrib notifications arriving early for Shia calculation methods
• •Prayer times now correctly use twilight angles for Jafari/Ithna-Asheri methods

Transcription Pipeline

• •Fixed audio cleaning returning empty files (service response handling)
• •Fixed word alignment returning zero segments (response parsing)
• •Fixed version history showing all entries as "Version 1"
• •Removed a legacy internal status check (not needed with cloud transcription)

Audio Verification Tool

• •Verify if audio was generated by Truth Promoters
• •Digital watermark detection confirms AI-generated audio
• •Metadata tag verification for attribution
• •Simple drag-and-drop interface for instant verification

Audio Attribution

• •All generated MP3s now include attribution metadata
• •Artist, album, title, and copyright tags embedded
• •Imperceptible digital watermark preserved in all audio

Centralized TTS Architecture

• •All text-to-speech now routes through the centralized orchestrator
• •Real-time progress streaming with job recovery support
• •Audio files now stored in cloud storage instead of temporary files

Enhanced Voice Cloning

• •Migrated to dedicated CPU server for better performance
• •Native paralinguistic tags: [laugh], [sigh], [cough], [chuckle], [gasp]
• •Model preloading for faster first request

Voice Clone Studio Improvements

• •Unified generate and history on single page
• •Title field for easy identification of generated audio
• •Pagination and sorting for history list
• •Full paralinguistic tags: [laugh], [chuckle], [sigh], [gasp], [cough], [sniff], [groan], [shush], [clear throat]
• •Pause buttons for natural speech pauses (0.5s, 1s, 2s, 3s)
• •ScribeFlow integration with URL pre-population
• •Voice Clone links on ScribeFlow artifacts (transcripts, translations, articles)

Ensemble ASR Settings

• •Enable/disable speech recognition engines from admin settings
• •Per-job engine selection for archival content transcription

Custom Notifications with Geo-Locking

• •Send targeted push notifications from the admin console
• •Geo-lock notifications by country and/or city
• •Choose notification icon from all publication logos
• •Notifications sent only to users who opted into community events
• •View eligible subscriber count before sending

Dual AI Provider Support

• •Added support for multiple AI providers with automatic failover
• •Configure primary and fallback models for guaranteed quality
• •Visual provider tags in the admin interface
• •View remaining credits from all providers

Enhanced AI Settings

• •Fallback model selector with enable/disable toggle
• •Active models summary with color-coded provider tags
• •Per-task configuration for Translation, Correction, and Article generation

Improved Archival Audio Processing

• •Better handling of low-quality archival recordings
• •Support for the latest generation of hardware
• •Improved memory management for large files

Artifact Version History

• •All transcriptions, translations, and articles are now versioned
• •View previous versions with date, time, and version number
• •One-click restore of any previous version
• •Preview content before restoring

System Status Dashboard

• •Live health monitoring of all services
• •Configurable auto-refresh interval
• •Color-coded status indicators for quick assessment

AI Settings Page

• •Choose translation and article generation models
• •Token pricing and cost estimates per operation
• •Test panel to validate connectivity
• •Encrypted API key storage

ScribeFlow - Transcription & Translation Pipeline

• •Upload audio/video files up to 2GB for transcription
• •Multi-language speech recognition for English, Arabic, Farsi, and Urdu
• •AI-powered translation to multiple languages
• •Automatic article generation from transcripts
• •Export to SRT subtitles and TXT transcripts
• •Real-time progress streaming for all operations