Changelog
v1.1.0 — 2026-10-20
A major update focused on the subtitle workshop and long-form quality.
Subtitle workshop
- ASS end to end: import
.assas a transcript; export ASS from Watch/Batch with built-in or custom style templates. - Explicit playlists: create, rename and delete lists in the library; play, reorder and remove items from the detail drawer.
- Beautify: Netflix profile (1–7s / 2-frame gap / 2 lines) with 24/25/30fps snapping; seven-rule report, preview up to eight changes, one-step undo.
- QC v2: nine rules including long gaps and out-of-bounds timing, with an issue histogram.
- macOS spellcheck: word-level suggestions; applying a replacement goes through undo.
Transcription and playback
- Honest Qwen capabilities: pseudo timing is labelled when no aligner is installed (burn-in/timed export allowed); native timing with an aligner.
- Live and long-form quality: live ASR self-healing, segment-boundary preservation, Whisper beam/best-of cap, faster retranscribe dialog.
- Player and transcript: opening transition, compact subtitle table, streaming transcript layout, artwork CSS fix.
Automation and platform
- MCP: tools grow from 2 to 5 with read-only list/read/model tools.
- Platform honesty: SenseVoice is hidden on Windows for now; system audio is disabled outside macOS; dual-source recording suggests headphones.
- Quality baseline: cargo 708/0/24, main chunk 1742.9 KB / gzip 545.0 KB.
v1.0.0 — 2026-06-02
Subwright is officially launched!
Core Features
- Offline transcription: Powered by OpenAI Whisper, fully local processing
- Multiple sources: Local files, microphone recording, system audio, YouTube links
- Live preview: Segments append during transcription
- Multiple exports: TXT, SRT, VTT, JSON, CSV, DOCX, Markdown, HTML, PDF
AI Capabilities
- Smart summaries: Auto-generate content summaries and key insights
- AI chat: Multi-turn Q&A based on full transcript
- Translation: 12 target languages supported
- AI polish: Remove fillers, correct grammar, smart paragraphs
- Mind maps: AI-generated hierarchical visualizations
- Smart chapters: Auto-identify topic transitions
Pro Features
- Speaker diarization: Up to 4 speakers, color-coded
- Subtitle burn-in: Custom font, position, color, outline
- Video trimming: Delete segments by text
- Audio enhancement: Noise reduction, vocal boost, phone audio optimization
- Batch processing: Queue and watch folder
- Cross-file analysis: Aggregate and compare batch results
System
- macOS 11+ / Windows 10+ support
- Apple Silicon / Intel dual architecture
- GPU acceleration (CoreML / Vulkan)
- 9-language UI
- Global hotkey dictation
- Auto-update
Last updated