Run the voice through a high-pass at 100 Hz, a downward expander, a de-esser, a 3 dB lift around 3.5 kHz, a compressor and a limiter, in that order.

Six-stage voice chain

A phone or clip-on recording sounds like a broadcast voice.

Use when: Every voice recorded outside a studio: creator video, a talking head, a podcast clip.

Prompt instructions

ProductAudience

Give me a sound brief: what to record or pick, where it sits on the timeline, the levels, and what to check before export, for a short video about [your product], for [your audience]. The technique is six-stage voice chain: a phone or clip-on recording sounds like a broadcast voice. How it works: run the voice through a high-pass at 100 Hz, a downward expander, a de-esser, a 3 dB lift around 3.5 kHz, a compressor and a limiter, in that order. It is right for: every voice recorded outside a studio: creator video, a talking head, a podcast clip. The expander goes before the compressor. Measured on the dry half of the source clip's own hook: the chain lowered the noise floor from -23.6 to -33.8 dB and kept the peaks under -1 dBFS. Be concrete: numbers, seconds and settings, no theory. Ask me one question first if something you need is missing.

The same prompt works in Claude, ChatGPT and Gemini.

Watch out

The expander goes before the compressor. Otherwise the compressor lifts the room noise between the words again.practitioner: Director Chair notes on Product UI as the explainer (Patchd)
Measured on the dry half of the source clip's own hook: the chain lowered the noise floor from -23.6 to -33.8 dB and kept the peaks under -1 dBFS. The values are starting points, not tuned to a room.measured: Director Chair notes on Product UI as the explainer (Patchd)
Set the expander's threshold above the room's own noise, measured on a second of silence. Loudness is a separate last step.practitioner: Director Chair field notes on editing

Notes

The six stages, in order

#	Stage	What it does
1	High-pass at about 100 Hz	removes rumble, where there is almost no voice
2	Downward expander	turns the quiet moments down further, so fans, keys and breaths drop between words
3	De-esser	turns down only the sharp hiss of S sounds
4	Lift of 3 dB around 3,500 Hz	the range where words get their clarity
5	Compressor	narrows the gap between quiet and loud words
6	Limiter	a ceiling, so a laugh or a shout does not distort

The video sits at the level of its neighbours in the feed and does not distort.

Loudness targetPrompt

A generated line stops sounding generated between the words.

Room tone under a generated voicePrompt

She says this mic has nothing on it, and in the middle of the sentence the sound changes to the processed version; below her, six empty numbered slots wait.

The result in the hookPrompt

Sources

Own words from Director Chair notes on Product UI as the explainer (Patchd) and Director Chair field notes on editing Checked 2026-11-13.

For agents: this card as markdown, as JSON.

Camera

Frame

Light and look

People and places

Motion

Edit

Story

Recipes

Tools