VoidCutter

VoidCutter is an open-source Adobe Premiere Pro UXP panel for transcript-driven cleanup, short-form clip discovery, and publishing metadata.

Official website · Download VoidCutter 1.0.0 (.ccx) · GitHub (primary) · GitLab mirror

GitHub is the primary repository. GitLab is maintained as an official backup mirror with the same published source history, tags, and release downloads.

VoidCutter can change a timeline. Keep Create a sequence copy before editing enabled until you have verified the result on your own material.

What it does

Tool	Purpose	Gemini required
Remove Silence	Finds transcript gaps and cuts or removes them using deterministic timing rules.	No
Remove Retakes	Suggests repeated takes, abandoned fragments, and removable recording chatter for review.	Yes
Create Shorts	Finds self-contained moments and creates one independent Premiere sequence per selected result.	Yes
Generate Metadata	Builds ordered descriptions, chapters, hashtags, tags, and unchanged Custom Text blocks.	Yes

Effects, Captions, audio-based silence detection, and automatic aspect-ratio reframing are visible future areas and are not implemented in version 1.0.0.

Premiere transcript required

Important

VoidCutter 1.0.0 does not transcribe audio itself. It reads the current transcript of the active Premiere sequence. Create or refresh that transcript before running a workflow.

1. Open the sequence you want to process and click its timeline.
2. Open Window > Text and select the Transcript tab.
3. Open the three-dot menu in the Transcript panel.
4. Choose Transcribe sequence. If a transcript already exists, choose Re-transcribe sequence instead.
5. Select the spoken language and any required transcription options, start transcription, and wait for Premiere to finish.
6. Confirm that the Text panel shows the transcript for the active sequence.

Captions are not required. For better AI results, correct important names and badly recognized phrases before analysis. See Adobe's Text-Based Editing guide or the detailed VoidCutter User Guide.

Whenever a sequence timeline changes, use Re-transcribe sequence before running another VoidCutter analysis on that sequence. This includes manual timeline edits and changes applied by Remove Silence or Remove Retakes. Do not reuse a transcript that describes the previous timeline state.

Requirements

• Adobe Premiere Pro 25.6 or newer.
• Adobe Creative Cloud Desktop for installing the .ccx package.
• A completed Premiere transcript for transcript-based workflows.
• Internet access and a personal Gemini API key for AI-powered tools.

Gemini availability, quotas, and pricing are controlled by Google. Create and manage keys in Google AI Studio.

Installation

1. Download VoidCutter-1.0.0.ccx from the official website. Release notes and source snapshots are available from GitHub or the GitLab mirror. Do not download the automatically generated Source code archives unless you want the project files.
2. Double-click the .ccx file.
3. Approve Adobe's warning for an independently distributed plugin.
4. Restart Premiere Pro if it was already open.
5. Open Window > UXP Plugins > VoidCutter.

Adobe documents this installation flow in Install a UXP plugin.

First-time setup

Remove Silence works without an API key. Remove Retakes, Create Shorts, and Generate Metadata require your own Gemini API key:

1. Complete the Premiere transcript steps above.
2. Open the official Google AI Studio API Keys page and sign in with a Google account.
3. Copy the key created for your default project, or choose Create API key and follow the project dialog.
4. Open the gear icon in VoidCutter and expand AI Provider.
5. Keep Provider set to Gemini, paste the key into API Key, and wait for Saved securely.
6. Refresh the Model list, select an available text model, and click Test Connection.
7. Open a tool, adjust its settings, and review generated ranges before applying them.

The key belongs to your Google account; VoidCutter does not include or share a common key. Provider quotas, regional availability, and billing are controlled by Google. Never publish the key in screenshots, logs, project files, or GitHub issues. Google's current key-management instructions are in Using Gemini API keys.

Which Gemini model should I choose?

Model	Recommended use
Gemini 3.5 Flash-Lite	Recommended starting point for VoidCutter 1.0.0. It is fast and cost-efficient for metadata and frequent analysis runs.
Gemini 3.6 Flash	Try for stronger Create Shorts or Remove Retakes analysis when it appears in your model list and your quota allows it.
Gemini 3.5 Flash	Stable, quality-focused alternative when 3.6 Flash is unavailable.

Prefer a stable text model. Avoid image, audio, Live, Preview, and Experimental variants for normal VoidCutter workflows. Google changes model availability and limits over time, so the model list shown for your key is authoritative. Check Google's current model guide and your project's active rate limits instead of assuming a fixed daily request count.

For every control and workflow, see the User Guide. If installation, transcripts, provider setup, or stale results cause trouble, see Troubleshooting.

Privacy

VoidCutter does not upload raw audio, raw video, Premiere project files, project or sequence names, or media paths. AI-powered tools send only the transcript data and task settings required for the selected operation to the configured Gemini endpoint. Remove Silence performs no AI request.

The Gemini API key is stored through UXP secure storage. Custom Text metadata blocks remain local and are inserted into the final output without being sent to Gemini. Read Security and privacy before using confidential transcripts.

Current limitations

• AI quality depends on the selected Gemini model and Premiere transcript accuracy.
• Create Shorts keeps the source aspect ratio; automatic vertical reframing is not implemented.
• Complex nested, multicam, speed-remapped, or heavily linked timelines may behave differently from ordinary talking-head sequences.
• A timeline operation can stop after partial work inside a generated copy. Inspect that copy before retrying from the original sequence.
• Premiere may restore a previously saved panel layout instead of the preferred 500 x 600 size.

Who VoidCutter is for

VoidCutter may be useful to editors looking for a free, open-source alternative for parts of a FireCut or AutoCut workflow, including pause removal, retake review, Shorts discovery, and publishing metadata. It is not currently a feature-for-feature replacement for either product.

Feedback and development

Use GitHub Issues for reproducible bugs and focused feature requests. If GitHub is unavailable, use GitLab Issues as a fallback. Never include API keys, private transcripts, provider responses, media paths, project files, or confidential production information.

The runtime uses plain HTML, CSS, and JavaScript with no bundler or installed runtime dependencies.

npm test
npm run build:release

On Windows, use npm.cmd test and npm.cmd run build:release if PowerShell blocks the unsigned npm shim. See Contributing before changing timeline edits, transcript handling, provider requests, or credential storage.

License

VoidCutter is available under the MIT License.

Support the project

If VoidCutter is useful to you and you would like to support its development, donations are welcome on the Solana network:

F8t5SH87L8gJNBicpmnCv9zeEYtiZ5J93Ugx5eTcYciD

Network: Solana. Verify the network and wallet address before sending; cryptocurrency transactions cannot be reversed.