From claude-transcription
Remove background noise from an audio file. Cloud providers (Auphonic default, ElevenLabs, Dolby.io) or local (DeepFilterNet ML, ffmpeg afftdn non-ML). Use when the user asks to denoise, clean up noise, remove hum/hiss, or enhance speech audio before transcription.
npx claudepluginhub danielrosehill/claude-code-plugins --plugin claude-transcriptionThis skill uses the workspace's default tool permissions.
Reduce background noise in a speech recording.
Guides Next.js Cache Components and Partial Prerendering (PPR): 'use cache' directives, cacheLife(), cacheTag(), revalidateTag() for caching, invalidation, static/dynamic optimization. Auto-activates on cacheComponents: true.
Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).
Share bugs, ideas, or general feedback.
Reduce background noise in a speech recording.
If the goal is transcription, denoising is usually not necessary. POC at https://github.com/danielrosehill/Crying-Baby-Audio-Scrub showed Whisper handles moderate background noise (e.g. baby crying) without preprocessing — the cleaned and uncleaned transcripts differed by only 5–6 word choices over 119 seconds.
Reach for this skill when:
For mild/moderate noise prepping for transcription, skip this skill — go directly
to preprocess-for-transcription and then transcribe.
Read ~/.config/claude-transcription/config.json for denoise_provider and prefer_local. If config missing, default to Auphonic (cheapest cloud option).
User may override per-invocation: "denoise with elevenlabs", "use deepfilternet", "local denoise".
Auphonic (default — ~$0.15/hr equivalent)
https://auphonic.com/api//simple/productions.json with input_file, preset, or direct algorithm flags (denoise=true)AUPHONIC_API_KEYElevenLabs Voice Isolator
https://api.elevenlabs.io/v1/audio-isolationELEVENLABS_API_KEYDolby.io Media Enhance
https://api.dolby.com/media/enhanceDOLBY_API_KEY and an input URL (upload to Dolby temp storage first)DeepFilterNet (ML, CPU-efficient)
uv tool install deepfilternet or pipx install deepfilternetdeepFilter <input.wav> -o <output.wav>ffmpeg afftdn (non-ML, instant)
ffmpeg -i input.wav -af afftdn=nf=-25 output.wavWrite to <source_stem>.denoised.<ext> in the same directory as source, unless user specifies a path.
/configure.ffmpeg missing → tell user to install it.