hyperframes/docs/studio/audio-effects.mdx

---
title: "Mix audio"
sidebarTitle: "Audio"
description: "In the browser editor: effects and presets on a track, music that sits under narration, groups, muting, and automation over time."
---

The browser editor carries a full mixer. Every audio track has an effect rack: filters that
decide which frequencies it may occupy, dynamics that decide how its level
behaves, character effects that change the shape of the waveform, and space
effects that put it somewhere.

Everything you do here is written into the composition's HTML. There is no
separate session file, so a mix survives being handed to someone else, committed,
or rendered on another machine.

## Effects and presets

### Open a track's effect rack

Select an audio clip on the timeline. The right panel shows an **Audio FX**
section for it, listing the chain from top to bottom in signal order and ending
in **Out to mix**. An empty track reads "No effects on this track."

Do not confuse it with the **Effects** section on the same panel — that one is
visual effects for pictures, and has nothing to do with sound.

You can also reach it without the panel: the **FX** button on an audio track's
timeline header opens a shelf of presets, with **Open rack ›** at the bottom. The
button shows a count when the track already has effects.

### Fix a voice with one preset

Presets are the fastest honest answer to "this voice sounds wrong". Each one is
an ordinary chain you could have built by hand — open it and the individual
effects are all showing.

Start from the symptom rather than the effect list:

| It sounds like | Reach for |
| --- | --- |
| Hum or thump underneath | `rumble-cut`, or a high-pass at 80 Hz |
| Boomy, chesty, too close to the mic | **Tame Boominess** |
| Muffled, like it is behind cardboard | **Reduce Mud** |
| Words hard to make out | **Add Clarity**, or [carve the music](/studio/audio-effects#make-music-sit-under-narration) |
| Harsh and tiring over a whole listen | **Soften Harshness** |
| Some words much louder than others | **Evenness** on a compressor, or **Even Out Levels** |
| Room tone audible between sentences | `room-gate` |
| Voice and music fighting each other | [A carve on the music](/studio/audio-effects#make-music-sit-under-narration) — not an EQ on either |
| Dry, recorded nowhere | `room-tight` or `room-natural` |
| Just "amateur" | `voice-clean` |

`voice-clean` is the default answer to "fix this voiceover": it removes rumble,
reduces mud, evens the level, adds clarity, and ends in a peak ceiling.
`voice-broadcast` is the same idea denser and more forward; `voice-warm` adds
body instead of cutting it.

Applying a preset **appends** to the chain, so cleaning a voice and then adding
character is a real thing to do. Re-applying the same preset replaces its own
nodes where they already sit.

Every named job also ships inside a preset at identical settings, so check what a
preset already contains before adding a job on top of it — `voice-clean` plus a
Reduce Mud job is −6 dB at 250 Hz where −3 was meant. Expanding the preset in the
rack shows its nodes by name, which is the fastest way to see it.

Beyond the corrective families there are **Character** presets — `telephone`,
`radio-am`, `megaphone`, `lofi-tape`, `pa-system`, `intercom`, `doofus-worble`,
`chipmunk`, `giant`, `monster` — and **Space** presets — `room-tight`, `room-natural`, `hall`, `slap-echo`,
`dub-throw`. Character presets are costumes rather than corrections, and they are
tuned to be distinguishable from one another, so do not stack two. `chipmunk`,
`giant`, and `monster` shift pitch, so they change who is speaking rather than
what they are speaking through.

### Add a single effect

Use **+ effect** in the rack. The menu groups effects the way you would reach for
them, in four families:

| Family | Decides |
| --- | --- |
| **Filters** | Which frequencies the track may occupy |
| **Dynamics** | How its level behaves over time |
| **Non-linear** | The shape of the waveform — character and grit |
| **Time** | Where it sits and how it moves: delay, reverb, chorus, phaser, and pitch shift |

The menu offers the [named jobs](#fix-a-voice-with-one-preset) in place of a bare
`peaking` filter, because picking `peaking` is picking a machine and leaving the
real decision — which frequency range — for afterwards.

Order matters, because each effect processes what the one before produced. Work in
this order and each step hears a cleaner signal than it otherwise would:

1. **Subtract before you add.** Cut rumble and mud first. A voice that sounds dull
   often has too much low-mid rather than too little top.
2. **Level after you filter.** A compressor reacts to whatever is loudest, and a
   rumble it can no longer see is a rumble it stops chasing.
3. **Relationships after level.** Carve the music once the voice itself is
   settled.
4. **Character, then ceiling.** Saturation and space go late; a limiter goes last,
   where it can actually act as a ceiling. Anything after it is not bounded by it.

Rename any node so the rack says what it is doing. Two `peaking` rows otherwise
look identical and you cannot tell the mud cut from the clarity lift.

Bypass a node instead of deleting it while you are deciding — it stays in the
chain, out of the signal path, so you can compare without losing the settings.

### Use the one-knob controls

Five effects have no single parameter that could honestly be their face — a
compressor's threshold means nothing without its ratio. Those show one derived
knob instead:

| Effect | Knob | 0 → 1 |
| --- | --- | --- |
| Compressor | **Evenness** | Barely touched → very even, quite squashed |
| Gate | **Tightness** | Only true silence → cuts quiet words too |
| Saturation | **Warmth** | Just a sheen → openly distorted |
| Reverb | **Space** | A small tight room → a big open hall |
| Bitcrush | **Crush** | Slightly gritty → destroyed |

Evenness, Warmth, and Space are level-matched: the make-up gain, output trim, and
dry leg move with the drive, so turning the knob up does not also turn the track
up. Tightness and Crush are not, because neither has a trim to move.

Editing a parameter underneath one of these knobs is allowed and simply moves the
knob — the chain stores the mechanism, not the knob position.

### Even out a drifting take

**Even Out Levels** measures the track's own speaking passages and writes a gain
envelope. Its target is that track's own 80th percentile rather than an absolute
level, so an already-even track is left alone. Use it when passages drift over a
whole take; use a compressor's **Evenness** when the problem is word-to-word
dynamics. **Remove levelling** takes it back off.

### What should happen

- Preview is what the render will sound like: both build the same graph from the
  same attributes.
- A chain the browser editor cannot read plays **dry** in preview so the composition stays
  workable — but it **fails the render** rather than shipping a track that sounds
  plausible and is wrong. If preview sounds unprocessed, suspect the chain.
- Reverb and delay make a track **longer** than its source. A bed with reverb no
  longer ends exactly at its clip length; that is expected.

### Common problems

**Nothing sounds different.** Check the node is not bypassed, then check you are
looking at **Audio FX** rather than the visual **Effects** section.

**The voice got louder as well as cleaner.** A preset ending in a peak ceiling
raises perceived loudness. Trim the track's own level rather than removing the
ceiling.

**A cut landed twice.** A preset already contained the job you added on top.
Expand the preset and read its nodes by name.

**It sounds worse after adding top end.** A muddy voice needs the low-mid cut
first; lifting the top of a muddy voice makes it muddy *and* harsh.

**There is hiss under the words.** Nothing here removes it. A gate closes the gaps
between phrases and leaves the noise beneath speech untouched — a source with
audible hiss needs a better source.

## Make music sit under narration

Music under narration is the most common audio problem in a video, and the
obvious fix is the wrong one.

Turning the whole music track down works, and it costs the music all of its
presence for as long as anyone is talking — the bed goes limp through the entire
voiceover. But a voice does not need the whole spectrum. It needs the few bands it
actually occupies. A **carve** takes only those out of the music, so the bed keeps
its low end and its top: still music, while the voice stays intelligible.

### When to use it

Whenever a music bed plays under speech. It is not a polish step to reach if
there is time — place both tracks, carve, and listen.

Skip it only when there is no narration for the music to sit under: a music video,
a title card, a montage cut to the track.

### Carve a bed

The carve is a module at the top of the **music** track's rack, not a separate
tool. Select the music clip, open its **Audio FX** section, and the module asks
you to pick the voices this bed should make room for.

A bed with exactly one candidate voice above it is **already carved**, at the
default strength, because that is what a bed under narration wants. Several
candidates leaves the picker waiting rather than guessing which one is the voice.

<Warning>
  The carve belongs on the **music**, never on the voice. It names the tracks it
  makes room for, the same way a sidechain compressor does — you select the track
  that gets quieter and pick what makes it quieter. A voice carved against itself
  is a bug, not a subtle mix choice.
</Warning>

### Set the strength

One control does the whole job. It derives how deep to cut, how many bands, how
wide they are, how far the level may drop, and how far under the voice to aim —
because those move together in any real mix.

| Strength | What you get |
| --- | --- |
| Default (0.25) | A 6 dB dip in three bands — audible without sounding like a hole |
| Around 0.5 | The dip reaches 10 dB, where a carve starts being heard as an effect |
| Higher | Deliberate territory for a loud bed under a quiet voice |
| 0 | Spectral only — one band, no level matching at all |

**Switch the carve off** keeps the settings and stops the carve, which is
different from clearing it: a bed with one candidate voice would otherwise be
carved again by default the moment it was re-evaluated.

### What it writes

Ordinary effects. Open the rack after carving and you will find a few `peaking`
filters and a `gain` stage — nothing hidden, nothing proprietary.

Two things follow from that:

- **The envelopes are editable.** The carve always follows the speech, so every
  value becomes an automation lane you can adjust by hand afterwards. Silence
  leaves the music alone; a loud passage pushes the carve to full depth. There is
  no fixed-depth mode, because a fixed dip thins the bed through every pause.
- **Re-carving is safe.** The nodes a carve writes are tagged as its own, so
  running it again at a new strength replaces exactly those and leaves every
  effect and lane you built by hand where it was.

Inside a carved bed the signal runs through the dips first, then the level match,
then anything you added yourself — which is why a limiter you place stays the last
ceiling.

### Point it at a group, not at clips

If narration arrives as several clips, [group them](/studio/audio-effects#group-tracks) and
carve against the group.

A carve pointed at individual clips has to list every one of them, and it stays
right only until the next edit: add a fourth narration clip and it plays outside
the carve's awareness, so the music silently fails to duck under it. A carve
pointed at a group resolves membership every time the analysis runs, so a clip
added later is covered without touching the carve.

The browser editor does this for you when you pick a second ungrouped voice — it creates a
group behind them and points the carve at that instead.

### What should happen

- The voice is legible without the bed sounding hollowed out.
- The music comes **back up between phrases** rather than staying flat.
- The bed keeps its bass and its top end. A carve that removed those would just
  be a filter.

### Common problems

**The music sounds notched or hollow.** Strength is too high. Come back toward
the default. This is the one failure mode with an obvious sound.

**The voice is still buried.** A carve cannot fix a bed that is simply louder than
the voice — spectral room does not solve a level problem. Compare the two tracks'
actual loudness first. Machine-generated speech commonly arrives far below a
mastered music track, and that gap is arithmetic rather than taste.

**The bed starts out already ducked.** A lane holds its first value backwards to
the start of its clip. A bed that begins before the voice needs an explicit "no
cut" point at its start.

**The music pumps.** The level envelope releases slowly on purpose, because music
that snaps back the instant a word ends sounds like a machine doing it. If it
still pumps, the gaps between phrases are shorter than the release — lower the
strength rather than fighting the envelope.

**Nothing was carved.** The picker leaves itself alone when several tracks could
be the voice. Name the voice explicitly.

## Group tracks

Five narration clips usually want one set of effects and one fader —
not five copies that drift apart as you edit. A group gives them that: its members
sum through a single bus, and the bus carries the group's own volume, effects, and
automation.

### When to use a group

- **Narration in several clips.** One rack for the whole voice, and one thing for
  a [carve](/studio/audio-effects#make-music-sit-under-narration) to point at.
- **More than a couple of sound effects.** One place to turn all of them down is
  worth more than precise individual levels you will never revisit.
- **Anything you want to turn up or down as a unit** — an ambience layer, a music stem.

A single clip does not need a group. Its own rack is already one place.

### Create a group

On an audio track holding more than one ungrouped clip, the **FX** button offers
to group them first: it reads "Group these clips to add effects to all of them".
Choose **Group**.

The browser editor creates a group behind those clips and points anything that needed it — a
carve, for instance — at the group rather than at the individual clip ids.

### Read a group row

A group gets its own timeline row, carrying:

| Control | Does |
| --- | --- |
| Caret | Shows or hides the member rows beneath it |
| `▤` and name | The group glyph and its label |
| Count | How many members it holds |
| **FX** | Opens the group's own rack, applying to the summed bus |
| `∿` | Shows or hides [the group's own automation lanes](/studio/audio-effects#automate-over-time) |

The caret and the `∿` toggle are independent: showing a group's members does not
open their automation lanes, and opening the group's lanes does not expand its
members.

A group-level effect hears the **sum** of its members, not each one separately.
That is the point of a bus — one compressor across a whole voiceover behaves
differently from five compressors on five clips, and usually better.

### Mute a track or a clip

On an audio track, the speaker button where a visual track has its eye mutes the
track: **Mute track** and **Unmute track**. For a selected clip, the same toggle sits
at the top of its properties, and the media section has a **Muted** switch.

A group row has no mute of its own; to silence a whole group, mute its members'
tracks. Muted audio is left out of the render, so unmute anything you want in the
export.

### Membership lives on the member

Worth knowing because it explains most surprises: each clip records which group it
belongs to, rather than the group listing its members. Delete a clip and it simply
disappears from the group; nothing dangles.

Groups are **audio only** and do not nest. A group marker on a video clip is
ignored.

### What should happen

- The group row appears with the members collapsed beneath it, and the
  grouping offer disappears from that track's **FX** button.
- Effects added to the group apply to everything under it at once.
- Muting the group silences all members, in preview and in the render.

### Common problems

**The group's effects do nothing.** Check that the members actually carry the
group, rather than that the group appears to list them.

**Nothing happened when I grouped clips.** Grouping resolves the clips it was
given against the timeline's current rows. If the track is collapsed, or its clips
live inside a nested composition, expand it and try again.

**Clips inside a nested composition are not offered grouping.** They arrive as
separate rows rather than as one track holding several clips, so the grouping
offer does not appear. Group them in the composition that owns them.

**A track was missing from the export.** Check whether it is muted: muted audio is
left out of the render.

## Automate over time

A static setting cannot say "quieter while the interview is on screen" or "fade
out over the last three seconds". An automation lane can: it is a set of points on
one parameter, and the value moves between them.

Lanes work on a track's own **volume** and on any supported effect parameter, and
a [group](/studio/audio-effects#group-tracks) has lanes of its own that apply to its whole bus.

### Show a track's lanes

Use the `∿` toggle on the track or group row. It shows a count when lanes already
exist, and it is independent of the row's expand caret — showing a group's members
does not open its lanes.

To start automating an effect parameter, automate it from the parameter's own
control in the rack. A parameter with a lane shows as **Automated**, and its
static control is disabled: the lane owns the value from then on.

### Draw and edit an envelope

Work directly on the lane:

- **Add a point** on the lane at the time you want it.
- **Drag a point** to move it in time or value.
- **Type an exact value** rather than dragging, when the number matters. A volume
  lane can go above unity — the ceiling is +12 dB, not 1.
- **Bend a segment** by dragging its curve, so the value holds and then moves late
  rather than travelling evenly.
- **Select a time range** to work on several points at once.

With a range selected, right-click it for the shape and cleanup menu:

| Item | Does |
| --- | --- |
| **Ramp up** / **Ramp down** | Replaces the selection with a single move in one direction |
| **Swell** | Rises and returns |
| **Dip** | Falls and returns — the shape most ducking wants |
| **Simplify** | Removes redundant points while keeping the shape. Needs at least three points in the selection |

A selection can also be **copied and pasted** onto another lane, and **retimed** by
stretching its edges — useful when narration moves and its ducking should move
with it.

### Two things that surprise people

**A track lane's times are relative to its clip.** A bed starting at 8 s has its
first point at composition time 8. Move the clip and its envelope moves with it.

**A group lane's times are composition time.** A group has no start of its own,
so its lanes are measured from the beginning of the video. If you are automating
a group bus, read the times as absolute.

**A lane holds its first value backwards** to the start of the clip, and its last
value forward to the end. So a bed that begins before the voice needs an explicit
"no cut" point at the start, or it begins already ducked. This is the single most
common automation mistake.

### Not every knob can move

A parameter can only be automated if the audio engine can schedule it directly.
Three kinds cannot:

| These | Cannot be automated because |
| --- | --- |
| Compressor, limiter, gate, bitcrush, pitch shift | They are configured whole rather than knob-by-knob — **no parameter on any of them can move** |
| Saturation's type, threshold, and oversample | They rebuild the distortion curve; only its output stage can move |
| Reverb's size and damping | They regenerate the room; its wet and dry levels can move |

**A lane on one of those is silently inert** — it stores fine and never does
anything. Pitch shift is the one that catches people out, because a rising pitch
sounds like something a lane should do: it cannot, and the lane will not report
it. To make a compressor's behaviour change over time, automate a `gain` stage
*before* it instead: that changes how hard the compressor is driven, which
is most of what automating its threshold would have done.

To fade a whole preset in or out, automate the preset itself rather than its
parts — its nodes share no single parameter, so the preset as a unit is the only
handle that works.

The [effect registry](/reference/audio-effects#effect-registry) marks every
parameter that can move.

### What should happen

- The value moves between your points, in preview and in the render identically.
- A parameter with a lane reads **Automated** and its static control is disabled.
- A [carve](/studio/audio-effects#make-music-sit-under-narration) writes lanes of its own, and they are
  ordinary lanes: you can edit them afterwards like any others.

### Common problems

**The lane does nothing.** Either the parameter is one of those above, or the
effect it pointed at is gone. A lane whose effect was deleted is dropped silently
rather than reported.

**A newly added effect arrives already "Automated".** A lane left behind by a
deleted effect can be inherited by the next effect that takes the same internal
id. Remove the stale lane.

**The volume lane is ignored.** A `volume` lane and a timeline tween on the same
track's volume conflict; the lane wins. Pick one.

**The envelope is right but the level is wrong.** A tween's values are absolute
and replace a track's own gain rather than scaling it. Check whether the track
also carries a static level.

**The bed ducks before anyone speaks.** The first point is holding backwards. Add
a point at the clip's start.

## Related topics

- [Audio effects implementation](/reference/audio-effects)