HN Lens

Deep Dive Chat Speech Synthesis (Read Aloud) Hands-Free Audio Reading: Added built-in Text-to-Speech controls directly beneath AI assistant responses in Deep Dive Chat. Neural Voice & Audio Ducking: Uses enhanced neural voice selection with automatic audio session ducking (.spokenAudio). Markdown-Safe Playback: Automatically strips Markdown formatting (like **, #, and URLs) before speaking so the audio playback sounds completely natural. Interactive Playback Controls: Play, pause, or switch playback between messages seamlessly. Audio automatically stops when navigating away from the chat screen. Thread-Aware Comment Analysis & Deep Discussions Nested Reply Fetching: Captures multi-level comment replies to surface substantive debates, technical counter-arguments, and dividing opinions. Hierarchical AI Prompts: Formats comment threads as structured parent-child debates for richer, more nuanced AI comment summaries. Dynamic Context Budgeting: Uses native Swift NaturalLanguage word guardrails to dynamically pack maximum comment context without overflowing model limits. Depth-First Hierarchy & Indentation: Displays comment replies directly under parent comments with visual indentation and thread markers. Overview Article & Story Detail Navigation Unified Navigation: Tapping any story card, digest summary, or category link in the Overview feed now seamlessly opens the full Story Detail view. State & Filter Preservation: Preserves your scroll position, active filters, and persona states when navigating back to the Overview section. Deep Dive Chat Formatting & Typography Strict Table Guardrails: Instructs on-device AI models to present structured information in clean bullet points rather than raw Markdown tables. Auto-Formatting & Spacing: Automatically inserts proper spacing before inline section headers and applies enhanced line spacing to AI response bubbles for better readability. Markdown Table Sanitizer: A post-processing fallback automatically converts residual table syntax into clean bullet lists.