Geel Media®

Your Timeline Deserves Better
By editors, for editors.

Developed in Costa Mesa, CA

Native AE Effects

Lumen
Analog
Riso
Flux
Drift
Prism
Origami
Fracture
all GPU Accelerated

Workflow

Torque (PP)
Ease (AE)
Xpress (AE)
Comp Importer (AE)

Free Tools

Sorter (AE + PP)
Layer Freeze (AE)
Render Collector (AE)
Glow Lite (AE)
Project Cleaner

Torque

Six tools. One panel. No subscription.

Torque is a Premiere Pro extension that handles the parts of editing you shouldn't have to think about — multicam switching, caption generation, silence removal, sequence resizing, project organization, and AI-powered social copy. Everything runs locally on your machine. No cloud processing, no monthly fee, no internet required.

Built by an editor who got tired of paying $30/month for tools that should've shipped with Premiere.

Multicam

Audio-driven camera switching that actually understands conversation.
Torque analyzes the audio on each camera track and builds a complete cut list based on who's talking. Assign your tracks as CAM (isolated mic) or WIDE (fallback shot), set your preferences, and hit Apply. The panel razor-cuts your timeline, disables the inactive cameras, and leaves you with a rough cut that's already 80% there.

It's not just "loudest mic wins." The system uses configurable reaction delay so cameras don't snap unnaturally fast, minimum hold times so quick interjections don't cause jarring flashes, and smart wide shot placement that snaps entry and exit points to natural speech pauses — not arbitrary timecodes.

Wide shots get inserted at regular intervals for visual variety, on speaker switches for coverage, and during silence for breathing room. Every insertion point is pause-aware: the cut lands where the speaker finishes a thought, not mid-sentence.

Track roles — CAM for isolated mics (analyzed), WIDE for your fallback camera (not analyzed). Up to 6 tracks.
Cutting Frequency — How aggressively the system commits to speaker switches. Low = conservative, high = reactive.
Reaction Delay — Shift all cuts forward or backward by frames. Positive feels natural. Negative anticipates the speaker.
Min Hold — Minimum time a camera stays on screen. Short interjections still cut, but they're padded so they don't flash.
Smart Wide Placement — Snaps wide shot boundaries to natural speech pauses within a configurable window.
Periodic Interval — Automatic wide shots every 15–90 seconds for visual variety.
Post-Wide Hold — After a wide shot, look ahead for the next speaker instead of defaulting back to the previous one.
Work on copy — Duplicates your sequence before editing. Your original timeline stays untouched.
Enable/Disable mode — Non-destructive by default. All clips stay on the timeline — inactive ones are just disabled.

Works with any number of speakers. RØDEcaster, Zoom, separate lav mics — if each speaker has their own audio track, Torque can sort it out.

Captions

Whisper-powered transcription with full style control — no third-party account needed.
Torque transcribes your timeline using OpenAI's Whisper model running locally on your machine. No API key, no upload, no per-minute billing. Pick a model size (tiny for drafts, small or medium for finals, large for multilingual), hit Transcribe, and the panel generates word-level timestamps from your audio.

Once transcribed, you get a full editable transcript with low-confidence words flagged in red and natural pause breaks marked. Click any word to correct it. Use Special Words to tag specific words with custom colors and fonts — brand names, guest names, key phrases — so they stand out in the final captions.

The style system gives you control over everything: font, size, color, bold, caps, punctuation, background bar, drop shadow, stroke, glow, and real-time word highlighting with separate "active" and "after" colors. The preview box updates live as you adjust settings, showing exactly how your captions will look on the timeline.

Position your captions at the top, center, or lower third with adjustable margins and broadcast-safe zone enforcement. Choose from five animation styles — Pop In, Slide Up, Stagger, Fade, and Typewriter — with adjustable speed.

Local Whisper transcription — No internet, no API, no per-minute cost. Runs entirely on your machine.
Model management — Download, switch, and delete Whisper models directly from the panel.
Transcript review — Low-confidence words flagged in red. Click to correct. Pause breaks shown inline.
Special Words — Tag words with custom color groups. Brand names, guest names, emphasis words get their own styling.
Preset system — Viral (large, bold, punchy) and Long-form (smaller, subtle) presets out of the box. Save your own.
Font management — Search and add any font installed on your system. Remove custom fonts you no longer need.
Drop Shadow / Stroke / Glow — Each with its own toggle, color picker, and fine-tuning controls.
Word Highlight — Highlights the currently spoken word in real time. Separate colors for "active" and "after" states.
Words per Group / Pause Break — Control how many words appear on screen together and how speech pauses affect grouping.
Hook text — Add a custom opening text overlay before captions begin. Full style control, separate from the main caption style.
Manual Text — Place styled text at the playhead position for callouts, labels, or lower thirds.
Live preview — See exactly how your captions will look before committing to the timeline.
5 animation styles — Pop In, Slide Up, Stagger, Fade, Typewriter. Adjustable speed.

Captions are built as Essential Graphics on your timeline — fully editable placement.

Silence Removal

Cut dead air in one click. Keep the breathing room you actually want.
Torque scans your audio track for sections that fall below a dB threshold for longer than a minimum duration, then removes them from your timeline. Simple as that.

The key is the buffer controls. Pre-silence padding keeps a bit of audio before each cut so you don't clip the tail end of a word. Post-silence padding keeps audio after each cut so the next word doesn't start abruptly. The difference between "choppy jump cut" and "tight, natural edit" is usually 100–150 milliseconds of padding.

Non-destructive mode disables silent clips instead of deleting them — easy to undo if you cut too aggressively. Destructive mode with ripple delete closes all gaps automatically, giving you a tightly packed timeline with no manual cleanup.

dB Threshold — Audio below this level is silence. Lower catches only dead silence. Higher catches quiet moments too.
Min Duration — A quiet section must last at least this long to count. Prevents cutting on natural pauses between words.
Pre/Post silence buffers — Keep 50–150ms of audio at each boundary so cuts sound natural, not chopped.
Non-destructive mode — Disables clips instead of deleting them. One undo brings everything back.
Destructive + Ripple — Deletes silence clips and closes all gaps. Timeline snaps together with no dead space.
Work on copy — Operates on a duplicate sequence so your original is always safe.
Full Sequence or In/Out — Process the entire timeline or just the marked range.

Works best on a mix track that captures all speakers.

Resize

Vertical, square, portrait — one click per sequence.
Load your project's sequences, check the ones you want to resize, pick a format, and hit Resize. Torque creates a new sequence at the target resolution for each selection, duplicates all your clips into it, and drops them in a [Resized] bin. Your original sequences stay completely untouched.

Seven built-in presets cover the formats that actually matter: 9:16 vertical for Reels/TikTok/Shorts, 4:5 portrait for Instagram feed, 1:1 square, and three landscape options up to 4K. Need something specific? Custom dimensions go up to 7680 × 4320.

Each resized sequence is fully independent — edits in the resized version don't affect the original, and vice versa. Frame rate is matched automatically.

Batch resize — Select multiple sequences and resize them all at once.
7 format presets — 9:16, 4:5, 1:1, 16:9 HD, 16:9 QHD, 16:9 4K, plus custom dimensions.
Non-destructive — Creates new sequences. Originals stay untouched in their original bin.
Auto-organized — Resized sequences go into a [Resized] bin so your project stays clean.
Search and filter — Find sequences by name when your project has dozens of timelines.
Frame rate matching — Output sequence inherits the source frame rate automatically.

After resizing, you'll want to reposition subjects for the new aspect ratio — Torque handles the sequence creation, you handle the creative framing.

Sorter

A clean project panel in two clicks.
Hit Scan and Torque reads every item in your project, categorizes it by file type — Video, Audio, Images, Graphics, Documents, Sequences, Linked Comps, Adjustment Layers — and shows you a preview of what will go where. Review the counts, then hit Sort. Everything moves into organized sub-bins under a parent folder you name.

Group by Extension adds another level: within each category, items are further sorted by file type (.mp4, .mov, .wav, .png, etc). Useful when you're working with mixed media from multiple sources.

Choose between sorting only loose items at the root level or reorganizing the entire project, including items already inside bins.

Scan before sorting — Preview exactly what will move and where before committing.
9 categories — Video, Audio, Images, Graphics, Documents, Sequences, Linked Comps, Adjustment Layers, Other.
Group by Extension — Optional sub-folders within each category (.mp4, .mov, .wav, etc).
Custom parent bin — Name the top-level folder whatever you want. Defaults to [Sorted].
Scope control — Root Level Only for quick cleanup, Entire Project for a full reorganization.
Accordion preview — Expand each category to see exactly which files will be sorted.

Social

Transcribe your edit. Generate captions for every platform. No copy-paste from ChatGPT.
The Social tab connects your timeline content to AI-powered caption writing. Select a sequence, transcribe it (using the same Whisper engine as the Captions tab), and generate platform-ready social media captions with one click.

Three AI providers are supported: Ollama for free local generation (no API key, no cost, no data leaves your machine), Claude API for Anthropic's models, and OpenAI API for GPT. Pick your provider, configure it once, and the panel remembers your setup.

The Ollama integration includes full model management — pull new models, see what's installed, delete ones you don't need, all from within the panel. No terminal required.

Generated captions come back as cards you can copy individually. Don't like one? Hit Regen for a fresh take. Toggle emojis, hashtags, and YouTube descriptions on or off before generating.

3 AI providers — Ollama (local, free), Claude API, OpenAI API. Use what you have.
Ollama model management — Pull, switch, and delete models without leaving Premiere.
Transcript-based generation — Captions are written from your actual content, not generic prompts.
Emojis / Hashtags / YT Description — Toggle each on or off before generating.
Per-sequence results — Each sequence gets its own caption card with Copy and Regen buttons.
Multi-sequence support — Transcribe and generate for multiple sequences in one session.
No data leaves your machine — With Ollama, everything runs locally. With API providers, only the transcript text is sent.

Ollama runs on most modern Macs and PCs. Recommended models: llama3 for quality, phi3 for speed on lower-spec machines.

Why Torque

One tool, not five.

Multicam editing, captions, silence removal, sequence resizing, project sorting, and AI social copy — six tools in a single panel. No juggling extensions from different vendors. No compatibility issues. One install, one interface, one place to learn.

No cloud. No upload. No wait.

Everything runs on your machine. Transcription is local Whisper — no API key, no per-minute billing, no uploading your unreleased episode to someone else's server. Audio analysis is stdlib and FFmpeg. Caption rendering is Essential Graphics on your timeline. Nothing phones home.

Audio-first multicam.

Torque analyzes each mic track and builds a cut list based on who's actually talking. It works with RØDEcaster, Zoom, lavs, booms, shotguns — any setup where speakers have their own track. Smart wide shot placement snaps cuts to natural speech pauses so they land where a thought ends, not mid-sentence. No face detection. No cloud AI. No guessing.

Captions that don't look like captions.

Full style control — font, size, color, drop shadow, stroke, glow, word highlighting, five animation styles. Tag specific words with custom colors for brand names and emphasis. Save presets so your look is locked in across episodes. Hook text for openers. Manual text for callouts. All editable on the timeline after placement.

Pay once.

One-time purchase. No subscription. No monthly fee. No per-minute transcription charges. No usage limits. Every update included. The tools run on your CPU — there's no reason to rent them.

Built by an editor.

Every feature exists because I needed it cutting my own show. Multicam switching that understands conversation pacing. Silence removal with padding controls so cuts don't sound chopped. Resizing that doesn't touch your original sequence. This isn't a feature factory — it's a tool built from actual production problems, by someone who still uses it every week.

What you'd pay elsewhere

AutoPod — $29/month ($348/year)
AutoCut — $20/month ($240/year)
Simon Says — $15/hour for transcription

Torque does multicam, captions, silence removal, resizing, sorting, and social caption generation — for a one-time price. The subscription model doesn't make sense for tools that run on your CPU.

Premiere Pro 2025–2026
macOS (Apple Silicon + Intel) · Windows