skills.rest
davdunc
davduncCOMMUNITY
Explore all skills by this author →
Website𝕏@davduncGitHub

AudioEditor

Generate word-level transcripts, classify segments, and apply ffmpeg edits with crossfades to clean audio recordings.

1|Updated Apr 16, 2026
Search Tags:#transcription#ffmpeg#audio editing#filler removal#podcast cleanup#cloud polishing
One-click install
npx skills add https://github.com/davdunc/pai-framework --skill audioeditor-davdunc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: AudioEditor
Source: https://github.com/davdunc/pai-framework/tree/main/skills/AudioEditor
Command: npx skills add https://github.com/davdunc/pai-framework --skill audioeditor-davdunc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill removes filler words, stutters, false starts, dead air, and verbal cut markers from recordings while preserving natural pacing and intelligibility.

Core Features & Use Cases

  • Word-level transcription: Creates timestamped transcripts using Whisper (insanely-fast-whisper on MPS when available).
  • LLM-based edit classification: Uses Claude to label segments as KEEP or specific cut types (CUT_FILLER, CUT_FALSE_START, CUT_EDIT_MARKER, CUT_STUTTER, CUT_DEAD_AIR, etc.).
  • Automated editing with smooth transitions: Applies 40ms qsin crossfades and fills gaps using room tone; attenuates breaths instead of removing them.
  • Optional cloud polishing: Runs a Cleanvoice API final pass for mouth sounds removal and loudness normalization (requires confirmation via the workflow design intent).
  • Preview and aggressive modes: Supports proposed-cut previews and tighter thresholds for more aggressive cleanup.

Quick Start

Run AudioEditor Clean on your recording to remove filler words, stutters, false starts, dead air, and cut markers.

Frequently Asked Questions about AudioEditor

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I remove filler words and stutters from a podcast recording?▼

Podcast cleanup works by generating word-level Whisper transcripts, classifying segments with Claude, and applying ffmpeg edits with 40ms crossfades to remove filler words, stutters, and dead air while preserving natural pacing.

What's the best way to automate audio cleanup for interview trimming?▼

Audio cleanup for interviews is best handled by combining Whisper word timestamps for segmentation, LLM classification for identifying false starts and edit markers, and ffmpeg crossfades to execute smooth cuts with room tone filling.

Can I preview audio edits before rendering the final podcast cleanup?▼

You can preview proposed cuts before final rendering, allowing you to evaluate classified audio segments for filler removal and stutter cuts before committing to the ffmpeg edit execution.

Does ffmpeg audio editing support aggressive threshold modes for dead air removal?▼

Aggressive threshold modes are supported for dead air removal, applying tighter cut parameters during classification while maintaining 40ms qsin crossfades and room tone filling for smooth ffmpeg transitions.

Do I need a Cleanvoice API key to remove mouth sounds during podcast cleanup?▼

A Cleanvoice API key is only required for optional cloud polishing like mouth sounds removal and loudness normalization; core filler removal and stutter cutting function using Whisper and ffmpeg without it.

Related Skills

Explore More Content & Communication Skills→
ruiyangruiyiruiyangruiyi

whisper

Transcribe spoken audio into timestamped text or English translations.

Community
Intermediate
BishopCodesBishopCodes
1

AudioEditor

Transcribe audio with word-level timestamps and remove filler words.

Community
Advanced
different-aidifferent-ai
249

video-subtitle-cutter

Transcribe video speech, detect filler words and pauses, and generate FFmpeg cut commands with SRT subtitles.

Official
Advanced
jykimjykim
49

video-cleaning

Remove long pauses and Korean filler words from video content using Whisper transcription and FFmpeg cuts.

Community
Advanced