builderauto-generated · unproven

How do I take a raw talking-head recording, transcribe it, remove filler words, hesitations and dead air to match a reference "good cut", and render a clean final MP4 with Remotion — replacing the manual editing agency

the agent that answers this

Raw recording → clean cut (Claude + Remotion). Take a raw talking-head recording, transcribe it, remove filler words, hesitations and dead air to match a reference "good cut", and render a clean final MP4 with Remotion — replacing the manual editing agency.

3 runsLast run 4d ago1 on a scheduleUpdated Jun 10v4

Cost: Free; on your own plan
Runs: On demand; or scheduled
Built from: 10 steps; plain language
Runs in: Claude or Codex; as you

what you get

A single clean MP4 with fillers/dead-air removed, a reviewable cut list (timestamps + words removed + runtime delta), saved back to the chosen destination after approval.

example result

illustrative

A finished run delivers: A single clean MP4 with fillers/dead-air removed, a reviewable cut list (timestamps + words removed + runtime delta), saved back to the chosen destination after approval.

Example output, to show the shape of the deliverable. It has not run on your data. Build the agent to get this on yours.

the steps

10 steps

Step 1tool
Locate the raw recording (and optional reference): local file path if given, else the founder's Drive editing folder via authenticated Chrome
Your model fills this step
- No integration? use a local file path
- No integration? open Drive in Chrome, read file IDs from DOM
Step 2tool
Get the raw into the working dir: skip if local, else gdown the public Drive link
Your model fills this step
- No integration? use local file in place
- No integration? gdown <fileId>
Step 3
Extract audio and transcribe to word-level timestamps with faster-whisper distil-large-v3; run ffmpeg silencedetect for pause/sound boundaries
Your model fills this step
Step 4
Establish the target editing style. Default for this user: remove fillers, fumbles, false starts, retakes (including camera-adjust repeats), AND non-speech sounds (coughs/clicks); KEEP all real content AND natural pauses up to ~3 seconds
Your model fills this step
Step 5
Build the keep/cut edit decision list: (a) remove filler words; (b) remove fumbles/false-starts/retakes including camera-adjust repeats, but do NOT over-cut a meaningful restate (only cut a repeat when the earlier take is abandoned or wrong); for back-to-back verbatim repeats extend the cut past the Whisper word-end since it under-runs; (c) cap dead air by INTER-WORD gaps (not just silencedetect, which leaves room-tone gaps): keep pauses up to ~3s, trim longer ones down to ~3s, trim silent intro/outro tight; (d) remove non-speech sounds (cough/click/breath) that show up as non-silent blips inside kept pauses; (e) resume each cut ~0.06s before the next kept word so its onset consonant is not clipped
Your model fills this step
Step 6decision
Review gate: present the cut list (timestamps, removals, runtime delta) for approval before rendering
Decision step
Step 7tool
Render frame-accurately by concatenating approved kept segments via ffmpeg trim/concat (hardware h264_videotoolbox), preserving source resolution/fps
Your model fills this step
- No integration? ffmpeg trim+concat filter_complex
- No integration? Remotion only for motion-graphics overlays
Step 8decision
Quality gate: re-transcribe the rendered output and adversarially verify every flub/retake/sound is gone, no content was clipped at the seams (spot-check resumed words), intentional content and natural pauses up to ~3s are intact, and A/V is synced; fix problems before delivery
Decision step
Step 9decision
Quality gate: re-transcribe output, verify flubs/retakes/sounds gone and NO words clipped at seams (count speech runs via silencedetect for suspected doublings, since fragment transcription hallucinates), pauses <=3s, A/V synced; fix before delivery
Decision step
Step 10decision
Delivery gate: preview and, only on approval, save to destination (browser upload capped at 10MB; for multi-GB use drag-drop or rclone)
Decision step

runs hands-free with

Claude for Chrome
pulls live data from sites that have no API (an MLS, Zillow, your CRM web UI) and clicks through web tasks for you, so the data-gathering steps run themselves instead of leaving placeholders to fill

Connect these and the agent gathers its own data and delivers on a schedule, instead of leaving you blanks to fill.

Build and run this on your own Claude or Codex, free

Install Implexa, then say build the Raw recording → clean cut (Claude + Remotion) agent and approve the schedule. It runs as you, on your real data, on the subscription you already pay for, and gets sharper each run. Your agent's memory is yours and travels with you across Claude, Codex, and whatever comes next. About 5 minutes to your first real run.

one command. signs you up + installs the plugin in claude code. free tier forever.

common questions

How do I take a raw talking-head recording, transcribe it, remove filler words, hesitations and dead air to match a reference "good cut", and render a clean final MP4 with Remotion — replacing the manual editing agency?

The Raw recording → clean cut (Claude + Remotion) agent. A single clean MP4 with fillers/dead-air removed, a reviewable cut list (timestamps + words removed + runtime delta), saved back to the chosen destination after approval.

Is the Raw recording → clean cut (Claude + Remotion) agent free?

Yes. It runs on the Claude or Codex subscription you already pay for, so there is no extra AI bill and no per-run charge. You can build and run unlimited agents on the free plan.

How often does the Raw recording → clean cut (Claude + Remotion) agent run?

You choose: run it on demand, or put it on a schedule (hourly, daily, weekly). Once scheduled it runs unattended, as you, on your own machine.

What does the Raw recording → clean cut (Claude + Remotion) agent need to run?

Install Implexa into your Claude or Codex, then connect Claude for Chrome so it can gather its own data and deliver hands-free. Implexa never touches your accounts or credentials.

Does the Raw recording → clean cut (Claude + Remotion) agent use my data? Is it private?

It runs as you, on your own machine, on your real data. The model runs inside your own Claude or Codex, so Implexa never sees your data, accounts, or credentials. Your agent's memory is yours and travels with you across Claude, Codex, and whatever comes next.

How do I build the Raw recording → clean cut (Claude + Remotion) agent?

Install Implexa into your Claude or Codex, then say "build the Raw recording → clean cut (Claude + Remotion) agent" and approve the schedule. Implexa assembles the 10 steps and it runs on its own. About 5 minutes to your first real run.

Can I change what the Raw recording → clean cut (Claude + Remotion) agent does?

Yes. Tell it what to change in plain language and it revises its steps; the next scheduled run uses the change, with no re-scheduling. Every change is versioned, and a run can even propose its own improvements.

changelog

v4Jun 10manual
Fine-tuned step 5/8 from review feedback: remove non-speech sounds (coughs), cap dead air by inter-word gaps, lead ~0.06s before resumed words to avoid clipping, do not over-cut meaningful restates, support fine word-level cuts
v3Jun 10manual
added 1 step
v2Jun 10manual
Refined cut spec: keep natural pauses up to ~3s (only trim longer), remove fillers + fumbles/retakes incl. camera-adjust repeats; support local source files; ffmpeg trim/concat render
v1Jun 9generated
auto-generated from "Take a raw talking-head recording, transcribe it, remove filler words, hesitatio"

Agents are alive: every change is a version, and a run can propose improvements that get reviewed and applied.