audioeditor_skill
- TypeScript
10.2k
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill danielmiessler/personal_ai_infrastructure --skill audioeditor- SKILL.md4.1 KB
Overview
This skill provides AI-powered audio and video cleaning workflows: fast word-level transcription, LLM edit classification, automated cuts with crossfades, and an optional cloud polish pass. It focuses on removing filler words, false starts, stutters, and dead air while preserving natural emphasis and breathing. The output is a cleaned MP3 or WAV ready for publishing.
How this skill works
On invocation the pipeline transcribes audio with a whisper backend to obtain word-level timestamps, then uses an LLM to classify each segment as keep or various cut types (filler, false start, stutter, dead air). An edit engine applies precise cuts using short crossfades, room-tone gap filling, and breath attenuation. Optionally, a cloud polishing step runs through a service to remove mouth noises and normalize loudness.
When to use it
- Preparing podcast episodes for release (remove ums, stutters, and dead air).
- Cleaning interview or lecture recordings for transcription and publishing.
- Trimming false starts and filler words from voiceovers and demos.
- Creating a preview of proposed edits before committing changes.
- Applying an aggressive cleanup pass with cloud polish for final master output.
Best practices
- Run the required voice notification before any processing so systems and stakeholders are informed.
- Preview edits first with the preview flag to review proposed cuts before applying them.
- Keep original recordings safe; edit outputs are new files (non-destructive workflow).
- Use aggressive thresholds only when natural repetitions or rhetorical emphasis are not needed.
- Provide clean recordings when possible—less background noise improves transcription and classification.
Example use cases
- Clean a multi-host podcast episode to remove filler words and tighten pacing.
- Preview and accept suggested edits for a recorded interview before exporting.
- Run an aggressive clean + polish to prepare a marketing voiceover for broadcast.
- Remove long dead-air gaps from a lecture recording while preserving pauses for emphasis.
FAQ
The core pipeline runs locally (transcribe, analyze, edit). Cleanvoice cloud polish is optional and requires an API key if used.
Can I review edits before they are applied?
Yes. Use the preview mode to generate an edits list without modifying the audio, review it, then run the full pipeline to apply changes.
What types of edits are performed automatically?
The system classifies and can remove filler words, false starts, stutters, edit markers, and dead air, and applies short crossfades, room-tone filling, and breath attenuation.
