v0.6.1

What's Changed

• Add Discord community links to localized READMEs by @MartinDelophy in #35

Full Changelog: v0.6.0...v0.6.1

Timeline Studio v0.6.0 — Mobile timeline and sticker workflows

Timeline Studio v0.6.0 makes mobile editing substantially more direct and predictable.

Highlights

• Touch-first mobile timeline with a fixed centered playhead and stable pinch zoom
• Track-specific mobile clip actions and direct audio/sticker property sheets
• Functional mobile track locks and safer gesture arbitration
• Pitch-preserving audio speed changes shared by timeline, preview, and export
• Reliable waveform placeholders and compact mobile duration labels
• Large-canvas close no longer clicks through to controls behind the modal
• Independent same-asset sticker instances with stable cross-subtrack placement
• Video-style sticker drag feedback with a ghosted source and time-accurate destination placeholder

Validation

• Full repository check passed
• 203 unit tests passed
• 11 targeted Playwright mobile/timeline tests passed
• Production build completed successfully

See PR #34 for the complete implementation details.

Timeline Studio v0.5.0

Timeline Studio v0.5.0

This release introduces a touch-first mobile editing workspace while preserving the established desktop editor experience.

Highlights

• Keeps the mobile project bar, preview, timeline toolbar, and option tabs fixed in one viewport
• Places the timeline directly beneath the preview with independent horizontal and vertical track scrolling
• Adds animated bottom sheets for Media, Captions, Smart, Audio, Stickers, and clip properties
• Uses denser 40px mobile timeline lanes while retaining the original 48px desktop lanes
• Adds a dedicated mobile Smart header and reduced-motion support
• Expands the Timeline Studio skill with browser workflow and end-to-end evaluation guidance

Validation

• 40 test files passed
• 154 tests passed
• TypeScript check passed
• Production Vite build passed
• GitHub CI quality check passed

Full changes: #31

Timeline Studio v0.4.1 — Editing guides and multilingual voices

Full Changelog: v0.4.0...v0.4.1

Timeline Studio v0.4.0 — AI Editing & Picture-in-Picture

English

Timeline Studio v0.4.0 makes the browser editor feel substantially closer to a modern desktop timeline while strengthening its local-first AI workflow.

Highlights

• CapCut-style picture-in-picture overlays with explicit drop intents, timeline placeholders, direct move/resize/rotate controls, and layer ordering.
• Clip-wide canvas transforms by default; keyframes are created only through explicit keyframe actions.
• Smoother native preview playback, kept separate from the deterministic offline export renderer.
• Reliable embedded source audio in exported files, separated-audio behavior, and localized offline-export status.
• New edit-timeline-studio Agent Skill for planning, executing, and verifying editable timeline projects.
• Focused English and Chinese documentation covering AI capabilities, Skills, startup, validation, and Netlify deployment.

Validation: typecheck, 133 unit tests, and production build passed.

中文

Timeline Studio v0.4.0 让浏览器剪辑体验更接近现代桌面时间线，同时继续强化本地优先的 AI 工作流。

重点更新

• 新增剪映/CapCut 风格画中画：明确区分“添加到主轨末尾”和“创建画中画”，支持轨道占位、画布移动/缩放/旋转与图层排序。
• 画布直接变换默认作用于整个片段，只有显式操作才创建关键帧。
• 使用更流畅的原生预览播放路径，并与确定性离线导出渲染器保持架构分离。
• 修复内嵌视频原声导出、分离音频行为，并完成离线导出状态国际化。
• 新增 edit-timeline-studio Agent Skill，用于规划、执行与验证可继续编辑的时间线项目。
• 重写中英文 README，聚焦 AI 能力、Skills、启动、验证与 Netlify 部署。

已通过类型检查、133 项单元测试和生产构建。

v0.3.4 — Multilingual voice reliability

Released July 17, 2026.

Highlights:

• Verified all nine listed voices across Chinese, English, German, Spanish, French, Italian, and Brazilian Portuguese.
• Unified Piper model persistence in OPFS and removed duplicate cache copies.
• Improved WebGPU adapter detection with reliable q8 WASM fallback.
• Fixed repeated voice generation after a generated audio clip is selected.
• Included the latest Auto Edit incremental-processing refinements.

Validation: 100 unit tests and all multilingual voice end-to-end cases passed, with successful typecheck, lint, and production build.

v0.3.3 — Auto Edit and offline workflow improvements

Released July 17, 2026.

Highlights:

• Browser-native Auto Edit with dynamic scene detection and candidate-frame review.
• Sliding-window local model analysis with incremental caption results.
• Improved caption timing normalization and one-to-one candidate output.
• Offline export, caption layout, sticker geometry, and timeline refinements.
• Unified model caching to prevent browser storage quota failures.

Validation: 100 tests passed, with successful typecheck, lint, and production build.

Timeline Studio v0.3.2 — Visual Editing & Smooth Timeline

Highlights

• Added clip-scoped Visuals editing with grouped and property-level transform keyframes.
• Added rectangle, rounded-rectangle, and circle masks with direct canvas manipulation, feathering, inversion, and preview/export parity.
• Added Basic, Mask, Speed, and Effects inspector tabs with video speed presets and fine adjustment.
• Added default-linked source-audio synchronization across preview, timeline geometry, project history, saved projects, and export, plus an explicit independent mode.
• Added pointer-anchored mouse-wheel timeline zoom over occupied clips, with lightweight gesture previews and same-frame scroll anchoring for smooth, jitter-free interaction.
• Expanded browser Piper voice coverage and editor-level multilingual voice validation.

Validation

• 28 targeted unit tests passed.
• Real-video Playwright end-to-end regression passed with ai-voiceover-16x9 (5).mp4.
• Production build passed.
• ESLint completed with 0 errors (existing warnings only).
• Netlify production deployment verified with no browser console errors.

Production

https://video-editor.ai-creator.top

Timeline Studio v0.3.1 — Hugging Face Model Hosting

Deployment optimization

• Moved the 66 MB vocal-separation model and all 16 TensorFlow.js weight shards to Hugging Face.
• Reduced the GitHub repository and Netlify production bundle size.
• Updated the browser model URL to the public Hugging Face model repository.
• Preserved service-worker caching for repeat inference without re-downloading model files.

Model: https://huggingface.co/haixin/timeline-studio-vocal-remover

Validation

• Production build passes.
• Timeline tests: 9/9 passing.

Timeline Studio v0.3.0 — Browser AI Vocal Separation

What's new

• Added fully browser-side AI vocal and instrumental separation.
• Bundled the UVR model and 16 weight shards for reliable local loading and service-worker caching.
• Added TensorFlow.js 4.22 WebGPU acceleration with automatic WebGL fallback.
• Added end-to-end validated STFT, model inference, inverse STFT, and stereo WAV output.
• Added Chinese and English copy for separation progress, errors, output names, and notifications.
• Kept hidden timeline clips visible with a disabled treatment instead of removing them.
• Added independent visibility state for multiple caption and voiceover lanes.
• Added the modern mobile web app capability metadata.

Validation

• Production build passes.
• Timeline tests: 9/9 passing.
• Browser E2E vocal separation produced valid vocal and instrumental WAV outputs.

Notes

The vocal-separation model is approximately 66 MB and is cached after its first load.