AVE 1.0.11

AVE 1.0.11 Release Notes

Status: Released on July 6, 2026 Bangkok time. The macOS app and DMG were signed, notarized, stapled, validated, tested from the DMG, and published to GitHub Releases.

Base: changes after 0fb4ed3 (Mark AVE 1.0.10 released) through 929d620 (Add timeline audio actions menu).

Highlights

• Added the new project format path with schemaVersion: 2, analysis sidecars, rebuildable project indexes, and downgrade-safe embedded analysis.
• Added global analysis reuse so re-importing footage that was already analyzed in another saved project can reuse transcript, visual, and local analysis instead of starting over.
• Added speech editing commands for removing silences, cutting long pauses, and removing filler words with approval-gated timeline edits.
• Kept generated captions linked to their source clips through close-gaps, trims, and speech-edit recuts, including karaoke word timing arrays.
• Added source-result preview support so footage results can be inspected before placing them on the timeline.
• Added speech-aware music ducking through AI actions and the toolbar audio actions menu.
• Added a toolbar audio actions menu for ducking, normalizing selected clips, voice cleanup, background-bed levels, smooth fades, and clearing volume automation.

Project Format And Reuse

• Projects save with the v2 format while keeping embedded analysis for downgrade safety.
• Sidecar and index write failures warn without blocking saves, and corrupt sidecars do not block project open.
• The project index remains disposable and rebuildable from project.json plus valid sidecars.
• Imported media can reuse analysis from the current project or recent saved projects when the source path or content fingerprint matches.

Timeline Editing

• Speech editing routes now run before generic trim/sequence routes so commands like "remove silences" and "remove ums and uhs" do not get stolen by unrelated planner paths.
• Caption-follow logic keeps generated captions aligned when clips move, close gaps, resize, or get recut by speech editing.
• Close gaps now treats video as the timing anchor and moves attached captions/graphics with their owning clips instead of flattening every track independently.

Audio

• Speech-aware ducking lowers music/audio beds under transcript-backed speech with generated volume keyframes.
• The toolbar audio menu exposes common human-facing audio actions without requiring chat.
• Ducking target selection avoids voice/interview tracks and protects audio-only speech clips unless a transcript-backed video speech source exists.

Verification

Release validation:

• npm run build
• npm --prefix render-worker run build
• node --check electron/main.cjs
• node --check electron/preload.cjs
• node --check scripts/assert-notarization-env.cjs
• node --check scripts/audit-desktop-package.cjs
• npm run ai:prompt-smoke
• npm run ai:prompt-regression
• npm run test:run
• npm audit --json
• npm --prefix render-worker audit --json
• git --no-pager diff --check
• env APPLE_KEYCHAIN_PROFILE=notarytool npm run desktop:dist
• npm run desktop:audit-package

Updater validation:

• release/latest-mac.yml exists and reports version: 1.0.11.
• release/AVE-1.0.11-mac-arm64.zip exists.
• release/AVE-1.0.11-mac-arm64.zip.blockmap exists.
• release/AVE-1.0.11-mac-arm64.dmg exists.
• release/AVE-1.0.11-mac-arm64.dmg.blockmap exists.
• release/mac-arm64/AVE.app/Contents/Resources/app-update.yml exists.
• The updater ZIP contains AVE.app/Contents/Resources/app-update.yml.
• Public latest-mac.yml resolves to version: 1.0.11.

Known Caveats

• The DMG install path was tested before publication.
• The updater artifacts and public metadata were verified. Do not claim the installed previous-version updater path passed unless separately tested from an installed older build.