Changelog

What's New

Refined Echoic interface

• Reworked the app-wide dark theme so the main window, menu bar, meeting views, dictation history, onboarding, and settings feel more consistent.
• Updated Echoic icon and logo assets across the app.
• Improved the main navigation and meeting workspace layout for better scanning during daily use.

Better meeting and AI workflows

• Polished meeting tabs, live transcript views, summaries, notes, scratchpad, audio playback, and imported meeting screens.
• Refined AI chat, source cards, provider settings, and AI template handling.
• Added coverage around AI templates, design system behavior, and meeting markdown export.

Improved

• Release metadata is now mirrored across both the public downloads repo and the private source repo.
• Beta and stable release workflows now use the public releases repo for DMG assets while keeping matching tags and release notes in the main repo.
• The release build now uses the expected Iterelle Apple Development signing identity and validates the Sparkle DMG signature metadata.

Fixed

• Fixed the release build's Moonshine package product reference so Xcode can resolve the current moonshine-swift package.
• Fixed release packaging docs and appcast examples that still pointed some download URLs at the private source repo.

Known Issues

• Apple Silicon only — Intel Macs are not supported in this beta.
• Not notarized yet — right-click -> Open may still be required the first time you launch the app.
• Some speech engines require a one-time model download before first use.

What's New

More reliable first launch

• Fixed an issue that could stop the app from opening cleanly in the beta build.
• Improved the download build so the installer is easier to prepare and verify.

Better transcription setup

• Fixed the Moonshine speech engine connection so the app can use it correctly.
• Updated the Parakeet speech engine support so meeting transcription works with the latest engine changes.

More dependable meetings

• Improved recurring meeting detection, so titles like “Standup” and “Daily Standup” are treated as the same series more reliably.
• Added extra checks around meeting import, search, summaries, and saved meeting data to catch regressions earlier.

Quality pass

• Cleaned up several beta blockers found during release testing.
• Verified the app build, test suite, installer, and app signature before publishing.

Known Issues

• Apple Silicon only — Intel Macs are not supported in this beta.
• Not notarized yet — right-click → Open may still be required the first time you launch the app.
• Some speech engines require a one-time model download before first use.

What's New

Core Dictation

• Multi-engine speech-to-text — choose from Parakeet v3 (recommended, ~542 MB), Whisper Large v3 Turbo (90+ languages, ~632 MB), Moonshine (fastest, ~288 MB), or Apple SpeechAnalyzer (zero download)
• Live text streaming with sub-500ms latency for all engines
• Floating dictation overlay — type into any app with a minimal pill bar UI
• Hold-to-record hotkey with push-to-talk and toggle modes
• Dictation history with searchable split-panel layout

Meeting Intelligence

• Dual-stream meeting capture — separate mic ("Me") and system audio ("Them") pipelines
• Live transcript rendered as chat bubbles with speaker diarization
• AI-powered summaries with 17 built-in templates across 5 categories
• Action item extraction via dedicated AI pipeline
• Custom AI template builder — create your own summary formats
• Single Flow Mode for non-diarized transcription

AI Providers

• 6 AI providers — Ollama (local), OpenAI, Anthropic, Groq, xAI, and Apple Intelligence
• OpenRouter, Gemini, DeepSeek support with multiple custom endpoints
• Per-feature AI routing for flexible provider assignment

Productivity

• Calendar integration — EventKit-backed calendar events in menu bar
• Audio file import — drag in recordings for transcription
• FTS5 full-text search across meeting history with date filtering
• Multi-format export — PDF, DOCX, and Markdown
• Meeting scratchpad for notes alongside transcripts

App Experience

• Guided onboarding with permission setup for mic, accessibility, calendar, and system audio
• Menu bar panel with vibrancy blur and quick access
• Unified main window with sidebar navigation
• Color theme picker with swatch cards
• Echoic brand identity across the app

Infrastructure

• Sparkle auto-update framework with beta/stable feed separation
• TelemetryDeck analytics with privacy toggle
• Discord feedback integration with screenshot attachments
• DMG installer with custom volume icon

Known Issues

• Apple Silicon only — Intel Macs not supported (Moonshine ONNX runtime limitation)
• Not notarized — right-click → Open required on first launch to bypass Gatekeeper
• Apple SpeechAnalyzer engine is a stub (macOS 26+ only)