Video Editor & Export

PatchWork has a built-in video editor for stitching generated clips into a final MP4. It's not a replacement for CapCut or Premiere, but for short-form pipeline work it covers the full last-mile: timeline assembly, silence trimming, voice changing, auto-captioning, and export.

Where it lives

Open PatchWork → left sidebar → CREATE → Video Editor. The Video Editor has its own project list (separate from your generation Projects). Create a new editor project to land on the canvas below.

1

2

3

4

5

6

7

1. **Project name** — click to rename. Determines the default export filename.
2. **File toolbar** — New, Save, Open, Close, **Export**, **Export All**. Export saves the current timeline; Export All saves every timeline in the project.
3. **Timeline tabs** — `Timeline 1`, `Timeline 2`, ... Each timeline exports to its own file. Tabs let you keep multiple cuts (different lengths, different platforms) in one project.
4. **Assets panel** — drag generated clips and images here. Filter by `All` / `Video` / `Image`.
5. **Preview** — 9:16 by default; aspect ratio configurable per timeline.
6. **Properties panel** — sub-tabs: `Properties` (transform/scale/position on a selected clip), `Silences`, `Audio`, `Captions`.
7. **Timeline tracks** — multiple tracks per timeline. Track 1 is typically the speaking-scene audio + video; later tracks hold B-roll overlays, captions, etc.

The rest of this page walks each of the editor's main features in order: timeline tabs, silences, audio + voice changing, captions, and export.

Timeline tabs — multiple cuts per project

The tab row above the timeline lets a single project hold multiple independent timelines. A + next to the current tab adds another timeline; you can rename each tab.

Why you'd want more than one timeline:

• Different durations — a 15s cut and a 30s cut from the same clips
• Per-platform versions — TikTok vs. Reels (different aspect ratio or AI Label Trick variant)
• A/B variants — same clips with different caption styles or pacing

Each timeline is its own export. The export filename is built from the timeline name — so renaming Timeline 1 to tiktok-15s produces an export named tiktok-15s.mp4 (see Export and file naming below).

Silences — auto-detect and remove dead audio

The Silences tab on the Properties panel scans your timeline's audio for stretches that fall below a configurable threshold, then lets you remove them in one pass.

1

2

3

4

5

1. **Scope** — `Selected Clip(s)` or `Entire Video (Track 1)`. Most of the time you want Entire Video.
2. **Define silence** — signal below `-42 dB` for more than `150 ms`. Both values are editable.
3. **Define audio** — signal above `-41 dB` for more than `300 ms`. The 1 dB gap between silence-ceiling and audio-floor avoids flicker around the boundary.
4. **Defaults work for most pipeline workflows.** Avatar voiceovers from Veo have consistent loudness; the defaults catch the natural pauses between sentences without trimming the breaths.
5. **Analyze** — scans the timeline and highlights the detected silence ranges. You then approve or refine before they're deleted from the timeline.

When to bump the thresholds:

• Long natural pauses you want to keep (storytelling beat, comedic timing) — raise the silence minimum from 150 ms to 400-500 ms
• Mic noise above -42 dB (room hum, fan noise) — raise the silence ceiling from -42 dB to -38 dB
• Whispered or breathy delivery — raise the audio floor or you'll over-trim

Audio — enhancement + voice changer

The Audio tab has two features stacked: clip enhancement and voice changing.

1

2

3

4

5

6

7

1. **Enhancement** — Resemble AI Nova SR. Denoises + upscales to 48 kHz studio quality. Useful when avatar VO sounds thin or has background hiss.
2. **Enhancement scope** — `Selected Clip(s)` or `All Clips (Track 1)`. **Enhance** runs the pass.
3. **Voice Changer scope** — `Selected Clip(s)` or `Entire Video`. Voice Changer replaces the original voice with a target voice while preserving timing + intonation.
4. **Provider** — `Chatterbox (fal.ai)` or `ElevenLabs`. Pick based on availability of the target voice + cost.
5. **Target Voice** — dropdown of voices you've uploaded. Use **+ Upload voice** to add a new voice sample (one short clip of the target speaker; the provider clones from there).
6. **Quality** — `Standard` or `HD`. HD costs more and takes longer; use for shipping content, Standard for tests.
7. **Convert** — runs the voice replacement. The original audio is preserved; the converted track replaces it on the timeline.

When to use Voice Changer:

• Avatar voice doesn't match the persona — too young / too old / wrong gender for the archetype
• Cloning a real person's voice for client-branded content (with their consent + a voice sample)
• Consistency across accounts — same target voice across multiple avatars' workflows for a unified brand sound

When NOT to use it:

• Veo's voice already sounds right — every Voice Changer pass is a cost + a quality risk
• You haven't validated the target voice — upload, test on one short clip, then commit to the whole video
• The avatar's lip-sync matters — voice swaps don't re-sync mouth movement; if lip-sync is critical, regenerate the video with a different Veo voice instead

Captions — auto-transcribe and style

The Captions tab handles auto-transcription, manual transcript editing, style selection, and export.

1

2

3

4

5

1. **TRANSCRIPTION → Auto-Caption** — runs speech-to-text against the timeline audio. Produces word-level timestamps so captions sync to delivery, not just sentence boundaries.
2. **TRANSCRIPT → Apply Text / Copy Text** — edit the auto-generated transcript by hand. **Apply Text** writes your edits back to the captions. **Copy Text** copies the full transcript for use elsewhere.
3. **EXPORT CAPTIONS → TXT / SRT / VTT** — export captions as a separate file. SRT and VTT are standard subtitle formats for platforms that accept uploaded caption files (YouTube, etc.).
4. **STYLE** — 10 preset caption styles: Classic, Subtitle, Karaoke, Pop, Minimal, Impact, Typewriter, Block, Neon, Slide Up. Pick based on platform + tone.
5. **BACKGROUND / POSITION / WORDS PER LINE** (scroll down) — toggle caption background fill, set padding + corner radius, place top/center/bottom, set words-per-line cap (default 5).

Style picker — quick mental map:

Style	Best for
Classic / Subtitle	Restrained, broadcast-style. Good for educational / clinical content.
Karaoke	Highlighted word reads as it's spoken. High engagement for hooks.
Pop / Impact	Bold, big, attention-grabbing. TikTok / Reels native.
Minimal	Tight, low-distraction. Good for premium / aesthetic content.
Typewriter	Builds character-by-character. Storytelling / confessional.
Block	Solid background block per caption. Maximum readability against busy footage.
Neon	Outlined glow text. Late-night / energetic tones.
Slide Up	Animated reveal from below. Modern punch.

Default for the pipeline's pop-style sales videos: Pop or Impact. For storytelling / confessional: Typewriter or Minimal.

Workflow for captions on a Veo-generated voiceover:

1. Click Auto-Caption — the editor transcribes with word-level timing
2. Spot-check the transcript — names, brand-specific terms, numbers often need manual fixes (Apply Text after editing)
3. Pick a STYLE
4. Set POSITION (usually bottom-third for short-form vertical)
5. Cap WORDS PER LINE at 4-5 so lines don't wrap awkwardly
6. (Optional) Export as .srt if the destination platform wants an uploaded caption file

Export and file naming

The Export buttons in the file toolbar produce MP4 files. Two flavours:

Button	What it does
Export	Renders the currently selected timeline to one MP4.
Export All	Renders every timeline in the project, one MP4 per timeline. Empty timelines are skipped.

The exported file's name comes from the Timeline name, not the project name. So:

Project: "salvora-en1-rhodiola-V2-1"
  Timeline 1: renamed to "tiktok-15s"  →  exports as  tiktok-15s.mp4
  Timeline 2: renamed to "reels-30s"   →  exports as  reels-30s.mp4
  Timeline 3: empty                    →  skipped

Rename your timelines BEFORE you export. Untouched timeline tabs export as Timeline 1.mp4, Timeline 2.mp4, etc. — usable but not meaningful when you have 20 cuts across your project library.

Status toasts at the bottom of the editor confirm what was exported and what was skipped:

Exporting 2 tab(s)...
Skipped "Timeline 3" (empty)
Export complete: 2/3 exported

Output format:

• Resolution — matches the preview aspect ratio (9:16 → 1080×1920 for short-form, 16:9 → 1920×1080)
• Codec — H.264 MP4
• Audio — included (voiceover + any voice-changed track + caption-baked-in if captions are visible at export time)

Files land in your local downloads folder by default.

The AI Label Trick — TikTok Shop content

For AI-generated content posted via TikTok Shop, add a 0.01-second flash of a dark frame (or an obviously AI-generated face) at the very end of the timeline. This triggers TikTok's "AI generated" platform label, which avoids content-violation flags for synthetic media.

How:

1. Add a black image or a generic AI-rendered face as an asset
2. Drop it on the timeline at the very end
3. Trim its duration to 0.01 seconds (just below 1 frame at 30fps — it flashes for one frame max)
4. Export

The flash is invisible to viewers at scroll speed but registers in TikTok's detection passes. Skip this for Amazon-keyword and Meta-Shop content — those channels don't penalize AI content the same way.

When PatchWork's editor isn't enough

You want	Use external editor
Advanced color grading per scene	CapCut / Premiere / DaVinci
Complex multi-track audio mixing	DaVinci / Premiere
Motion graphics, animated overlays, custom transitions	After Effects / Premiere
Long-form content (>90 seconds)	DaVinci / Premiere
Team workflow already on a different editor	Whatever the team uses

PatchWork's editor is purpose-built for short-form AI-generated content. For anything more elaborate, external tools win on flexibility — but you lose the integration with workflow outputs and the AI Label Trick template.

When you're ready

→ Next: Chapter 5 — Quality, Testing & Prompt Tuning. Now that you can drive PatchWork's creative tools end-to-end, learn how to judge a generation and when to reach for the prompt-tuning skill.